u.

«A

 

 

 

:

" ﬁtn- 41A.”

,urnu-
, N,»

if Em;

“(4&9 - .-. -\
:W

37???

, . I. u!"
u J
11,;

‘3
.,
3m

4‘. ’
pf
pa”

, wax“ 2....“
5&5,

an-

"1 ~
. 1‘." , '1“.
' ’{m X?" :71 "
.4 i; E?!“ yup: - 5:1: "‘
at ».,Lt.. .3‘ -
' " ‘ mgr-$3; 1:“:
W“: :73ka

1:: ’5 '

..
..
3?". ﬂ

2:35; 2;

4:? vs;
14491: ’45:.

:1.
, 23:23:": 45...“: ~-
. {-13.13% .x ﬁ'f:
~96 M -
4“. i; -" -.
‘ x ' 3.31“ .§:
.us. «:5

 

1 r
, "1-9:, W?“ “I,“ ...
.931? 7:, , . g >- > "I" I—
'4‘ u ”in”? §
‘01-

 

 

 

 

‘13:, a... v; xﬁu
II ~v.l "I '1‘
v39: ‘ mi: 3%
: 4.5.? r. «51‘ 3 .‘ul‘

3“?
“cantféi ‘
. ,
-

4%
my I J» .1-
.uma‘u '-

-v~r
xiv/2E,"
4'2 ‘

fur-3'1? r-

w; ";‘3, . w“ ”at...” ”my

5.? www;

l 3 .~ 'l‘
2. 1yvu(" ‘ ' 47 ‘ '

.I‘ML. 13' "
,
1.1.3::
'1
,.. .,
3‘. 3:5 -"‘:. '
3 1.?

"I V“

M! "'1:L'.:-v
‘._ £13,; “ﬁx 1:
r.
1 71.3 1153'} 7
m... ,, .2.
“mm-43‘:
f“ '

:M—oﬂ‘.

.,. r

:‘tnrﬁ 43'?

v 'v' u _ ' ’
3 ,3 . - WW7”?! ”Mari" J3
I ». v,"- ' ‘ #1.” 73:7 Illa!“ ”En:o:‘5r -¢-
..,,.-.,. ; '-'. p 2., J 1)., :97. 7 my”. :1.

-.....~:_‘..,
»riI-Vm ’3: v: :,..-..4
"rt 0—-
m-

r:
..... m... -..
wmwc .

*-

...
“ ~m—mq3

..........

.. ”a...
.... ..
..

 

we:
W...
M‘

\ES

hi

I”?
\\

\iiiiiiiiiim

\

This is to certify that the

thesis entitled
DESIGN AND CHARACTERIZATION OF A

FLOW REVERSAL/FLOW RECYCLE
CONTINUOUS FLOW WYSIS SYSTEM
presen e y

Edwin B. Townsend IV

has been accepted towards fulﬁllment
of the requirements for

hIS d egree in Chemi 5 try

W/m

Major professor

 

Date j/aq/w

0-7539 MS U is an Afﬁrmative Action/Equal Opportunity Institution

' . 1.9

Univer first

5:;
(3
SJ"
‘5 _,
3‘33

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

—T| J

MSU Ie An Afﬁrmative Action/Equal Opportunlty Institution
cmma-M

 

 

DESIGN AND CHARACTERIZATION OF A FLOW REVERSAL/FLOW
RECYCLE CONTINUOUS FLOW ANALYSIS SYSTEM

by

EDWIN B. TOWNSEND IV

A DISSERTATION

Subnﬁﬂﬁdtn
Michigan State University
in partial fulﬁllment of the requirements
for the degree of

MASTER OF SCIENCE

Department of Chemistry

1991

Abstract

DESIGN AND CHARACTERIZATION OF A FLOW REVERSAL/FLOW
RECYCLE CONTINUOUS FLOW ANALYSIS SYSTEM

by

Edwin B. Townsend IV

A continuous ﬂow analysis system that is capable of performing either
ﬂow reversals or ﬂow recycles has been constructed. The key components are
two 6 port switching valves, which are used to allow the ﬂowing analytical
stream to be either reversed or recycled following the passage of sample
through the flow cell and the subsequent detection of a peak by the detector.
An interface has been developed that allows for complete computer control of
both the valve switching and the data acquisition after the operator enters
experimental parameters.

The new system is characterized through the use of a series of
photometric determinations. Both peak area and peak height are used to
characterize the sample peaks, and the variation in those measurements
caused by the number of reversals or recycles is determined to be minimal.
The analytical utility of ﬂow reversals and ﬂow recycles is demonstrated by
the generation of standard curves and the determination of nitrite. Finally, a
slow kinetic determination is chosen and compared with an alternate
photometric detection method to demonstrate the usefulness of reversals

and/or recycles.

DEDICATION

I'd like to take this opportunity to dedicate this thesis to my brother,
Kevin. He is currently serving aboard the USS. Biddle CG—34, and was in
the Red Sea during the Gulf War. He is brave enough to be doing
something that I could never do, and I have the utter most respect and
admiration for him. Beef, God Bless you and I'm very thankful that you

came home safely.

ACKNOWLEDGEMENTS

I would like to thank Professor Stanley Couch for his guidance,
friendship, and faith in me during the course of this research. Without his
belief in me, I would probably not be writing this thesis right now.

I would also like to thank my wife, Maria, for her support and
unbridled faith in me. Without her, my life would just be on one even,
unexciting, uneventful keel.

Many more thanks to the members of the Crouch group. Kris, SV M,
BQ, Odie, DT, Larry, and Dave S. all make lab a great place to work. I hope
that I have been able to make it the same for them. Also, thanks to our
"honorary" group member and resident QuickBasic expert, Eric
Hemenway. Without his help, the computer code might not have ever
gotten oﬂ‘ the ground.

I'd also like to thank all the others in "group". They include Knurt,
Tracy, Gary, Gilbey, Berly, Laura, Jason, Sue, Mary, Mark, Drs. Light &
Wahl-Light, and Dave G., among others. You guys make grad school a

real experience.

iv

TABLE OF CONTENTS

Page
List of Tables ............................................................................ VII
List of Figures ........................................................................... VIII
Chapter 1. Historical, Theoretical, and Literature Introduction
to Continuous Flow Analysis ..................................... 1
A. Overview ........................................................ 1
B. History of CFA ................................................ 2
C. Theory of Continuous Flow Analysis ................. 6
D. Comparable Flow Reversal and Flow Recycle
Systems .................................................. 16
Chapter 2. Instrumentation, Software, and Data Processing
Techniques .............................................................. 21
A. Overview ........................................................ 21
B. Set-up ............................................................. 21
C. Components .................................................... 3)
D. Sample Introduction ........................................ 37
E. Data Acquisition and Processing ....................... 37
Chapter 3. Characterization of the Flow Reversal/Flow Recycle
System .................................................................... 46
A. Overview ........................................................ 46
B. Experimental .................................................. 46
C. Instrumentation ............................................. 48

D. Peak Shape Determinations .............................. 48

1. Introduction .......................................... 48

2. Inﬂuence of Valve Switching ................... 50

E. The Griess Reaction ........................................ 55
1. Introduction .......................................... 55

2 Calibration Curves ................................. 57

3. Percent Interaction or Carryover .............. 59

4 Evaluation of Peak Detection Routine ........ 61
Chapter 4. Analysis of the Glucose Oxidase/Trinder Reaction ....... 63
A. Overview ........................................................ 63
B. Introduction ................................................... 63
C. Experimental .................................................. 64
D. Instrumentation ............................................. 68
E. Results ........................................................... $
F. Summary ....................................................... 76
Chapter 5. Future Projects ........................................................ 77
A. Overview ........................................................ 77
B. Valve Design ................................................... 77
C. Computer Enhancements ................................. 78
D. Segmental FIA ................................................ 84
E. Kinetics .......................................................... 85
F. Diode Array Detection ...................................... 85
G. Light Source Modiﬁcations ............... ' ................ 86
H. Other Possibilities ............................................ 86

Appendix A. Some Sample Output from the Flow
Reversal/Recycle Instrument ................................. 87

Appendix B. Some Sample Data Sheets from the Flow
Reversal/Recycle Instrument ................................. 91

List of References ....................................................................... %

LIST OF TABLES

Table Page

2-1 Flow rates for various pump tubes at the nominal
pump speed setting of 44 ..................................................... 3)

3-1 A comparison of calculated and experimental dispersion
values. All dispersion values are in seconds ........................ 51

3-2 A comparison of % relative standard deviations for 2.5 ppm
nitrite solution using CFA and ﬂow reversal/ﬂow recycle
CFA. The CFA data was obtained using three trials ............. 57

3—3 The ﬁgures of merit from the calibration plots of CFA, Flow
Recycle, and Flow Reversal using the Griess reaction.
Calibration plots are for both the peak area and the peak

height .............................................................................. 58
3-4 The % Interaction for nitrite samples with 4 recycles ............. 60
3-5 The % Interaction for nitrite samples with 4 reversals ........... 61

4-1 A comparison of the relative rates of glucose oxidase/Trinder
reaction. The data reﬂect the linear regression ﬁt to the rate
versus concentration data ................................................... 71

B-l Data sheet on experiments with 2 dye samples and

2 reversals ........................................................................ 92
B-2 Data sheet on trials with 3 dye samples and 1 reversal ........... 93
BB Data sheet for trials with 2 dye samples and 2 recycles ........... 94
B4 Data sheets for trials with 3 dye samples and 1 recycle ........... 9'5

Figure
1-1
1-2

1-33

1-3b

1-4

2—1

2—2

2-3

2-5

LIST OF FIGURES

Page
A generalized diagram of a typical CFA set-up ................. 7
(a) The ideal, ﬂat-topped, CFA peak. (b) The actual,
experimental CFA peak ................................................. 9
The effects of segmentation frequency, ﬂow rate, and
internal diameter of the ﬂow system on dispersion.
F=0.0083 mL 3'1, other variables are as in Ref. 27 .............. 14
The effects of segmentation frequency, ﬂow rate, and
internal diameter of the ﬂow system on dispersion.
For this plot, dt=0.1 cm. All other variables are as found
in Ref. 27 ...................................................................... 15

(a) A typical CFA set up with Air, Reagent, and Sample.
(b) A typical FIA set up where the sample is merged in the
reagent stream in a nonreactive carrier stream ............... 17

The general schematic for the Flow Reversal Continuous
Flow system ................................................................. 23

The general schematic for the Flow Recycle Continuous
Flow system ................................................................. 24

The instrumental arrangement of the Reversal/Recycle
system ......................................................................... 25

The valve conﬁguration for ﬂow reversal CFA with a

4-port and a 6-port valve. 3) shows the initial position

while b) shows the valve conﬁguration after a reversal.

Notice that in the reversal conﬁguration, valve B is

stationary. Note: All arrows exiting valves go to waste ..... %

The valve conﬁguration for ﬂow reversal CFA with two six
port valves. a) shows the initial position while b) shows the
valve conﬁguration after a reversal. Notice that in the
reversal conﬁguration, valve B is stationary. Note: All
arrows exiting valves go to waste .................................... 27

Figure

2—7

2-9

2-10

2-11

2-12

31

3-2

3-3

Page

The valve conﬁguration for ﬂow recycle CFA with a

4-port and a 6-port valve. 8) shows the initial position

while b) shows the valve conﬁguration after a recycle.

Notice that in the recycle conﬁguration, both valves turn.
Note: All arrows exiting valves go to waste as

The valve conﬁguration for ﬂow recycle CFA with two six

port valves. a) shows the initial position while b) shows the
valve conﬁguration after a recycle. Notice that in the recycle
conﬁguration, both valves turn. Note: All arrows exiting
valves go to waste .......................................................... 29

A plot of pump tube delivery factory versus pump setting,
indicating that linear pump motor speed adjustment is
possible. The delivery factor 1s the value by which the

nominal pump tube ﬂow rate must be multiplied to

estimate actual ﬂow rate ................................................ 31

A schematic diagram of the double-beam, ﬁlter photometer
used in this CFA system ................................................ 34

The "thought process" of the CFA Flow Reversal/Recycle
program ...................................................................... 38

An example screen that might be seen by the user during
an experimental run ...................................................... 42

The threshold value is determined using the difference
between the ﬁrst three points and the last three points in

a 10 point series. When the threshold is large enough, the
beginning of the peak is "determined". The end of the peak

is determined in the same manner, only the difference must
also be near the initial value used when determining the

start of the peak. The "x" represents the ﬁrst three points

and they" " represents the last three points in a series of 10
points .......................................................................... 44

The pump conﬁguration and tubing selections for the
manifold used to study the effect of valve switching on peak
shape using a phenol red/borax buffer system .................. 49

The change in peak area as a function of average peak area
after each successive reversal or recycle .......................... 53

The change in peak height as a function of average peak
height after each successive reversal or recycle ................ 54

ix

Figure

3-5

4—1

4-2

4-3

51

5-2

5-3

Page

The reaction scheme for the determination of N02“ using
the Griess Reaction ....................................................... 56

The pump conﬁguration and tubing sizes for the manifold
used for the Griess reaction ........................................... 57

A plot representing the frequency of time from the
endpoint detection of one peak and the detection of the
front of the peak following a valve switch ........... 62

The glucose oxidase reaction and the Trinder detection
reaction ....................................................................... 66

The reaction manifold and pump tubing choices for the
glucose oxidase/Trinder reaction .................................... 67

The rate of the glucose oxidase reaction using a single
beam, Heath instrument ................................................ 70

An example of the data obtained from glucose
oxidase/Trinder reaction using ﬂow reversals ................... 73

An example of the data obtained from glucose
oxidase/Trinder reaction using ﬂow recycles .................. . 74

The resulting absorbance versus time plot from the recycle
data in Figure 4-4 .......................................................... 75

The general schematic of the Flow Reversal/Flow Recycle
continuous ﬂow system. In this conﬁguration, connections
do not have to be switched when choosing between ﬂow
reversal or ﬂow recycle modes ........................................ 79

The valve conﬁguration for the ﬂow reversal conﬁguration

of the Reversal/Recycle valve design. a) shows the initial
position while b) shows the valve conﬁguration after a
reversal. Notice that in the reversal conﬁguration, valves B
and C are stationary. Note: All arrows exiting valves go to
waste ........................................................................... so

The valve conﬁguration for the ﬂow recycle conﬁguration

of the Reversal/Recycle valve design. a) shows the initial
position while b) shows the valve conﬁguration after a

recycle. Notice that in the recycle conﬁguration, all valves

are turned. Note: All arrows exiting valves go to waste 81

Figure

A—l

A-3

A-4

Page
The output from 1 dye sample with 5 reversals .................. 87
The output from 1 phenol red dye sample and 4 recycles ..... 88

The interaction test pattern for 2.5, 20.0, and 2.5 uM nitrite
solutions and 4 reversals ................................................ $

The interaction test pattern for 2.5, 20.0, and 2.5 uM nitrite
and 4 recycles ................................................................ £1)

Chapter 1

Historical, Theoretical, and Literature Introduction
to Continuous Flow Analysis

A. Overview
Flow analysis can roughly be deﬁned as the transport of a sample
from the start of sampling until reaching waste (1). In between the start
and the ﬁnish, several events may occur. These events include, but are not
restricted to, mixing, reacting, separating, and/or detecting one or more
analytes in the sample. Two well-known subdivisions of ﬂow analysis are
Air-segmented Continuous Flow Analysis (CFA) and Flow Injection
Analysis (FIA). Air-segmented continuous ﬂow analysis was ﬁrst
conceptualized by Leonard Skeggs (2,3) in 1957 and then developed
commercially by the Technicon Corporation as the AutoAnalyzer
(Technicon Instruments, Tarrytown, NY). Flow Injection Analysis was
developed nearly simultaneously in the 1970's by the pair of Jaromir
Ruzicka and E10 Hansen (4) in Denmark and by Kent Stewart (5) in the
United States. The major physical quality differentiating FIA from CFA is
the air bubble that is introduced into the ﬂowing stream in CFA, while FIA
is thought of as unsegmented or nonsegmented continuous ﬂow analysis.
This physical difference gives each system unique properties, as well as
offering advantages and disadvantages over the other.
Since this investigation deals primarily with continuous ﬂow systems,

only historical information, theory, and pertinent applications for CFA are

2

discussed. However, necessary FIA information and comparisons are

presented whenever apprOpriate.

B. History of Continuous Flow Analysis

1. Introduction

Air-segmented continuous ﬂow analysis was invented by Skeggs (2)
and commercially marketed by the Technicon Corporation. In an effort to
automate batch analysis, Skeggs, in the early 1950's, decided to forgo the
conventional idea of developing an instrument that included traditional
analytical tools such as test tubes and pipets, and instead to began working
on a novel idea to perform sequential batch analysis based upon a
continuously ﬂowing stream. This continuously ﬂowing stream would be
contained in interconnected, small diameter tubes called a manifold.
Initially, CFA was developed to satisfy the demands for more clinical
laboratory determinations, and in fact, Skeggs found CFA useful for such
things as removing interferences in clinical assays or for improving
clinical dialysis. However, his most signiﬁcant development in CFA was
"the bubble". Skeggs had found that the performance of his CFA system
was being signiﬁcantly handicapped because of the intermixing of samples
as they passed through the reaction manifold. He solved this intermixing
problems by inserting air bubbles, at regular intervals, into the ﬂowing
stream.

Thus, Air-segmented Continuous Flow analysis was born. Because
a very detailed account of the development of CFA by Skeggs and the
resulting commercialization of CFA by the Technicon Corporation is given

in a recent article (6), I will not go into great detail about every

3

development, but instead will highlight those which signiﬁcantly increased
the capability, performance, or versatility of CFA systems.

2. The AutoAnalyzer I

The ﬁrst patent in CFA was issued to Skeggs in 1957 (3). That same
year, Technicon sold a single-channel CFA system known as the
AutoAnalyzer (AAI). This system ﬁrst sold in 1957 differs little in basic
components when compared to the current systems of today. The
differences that do exist are due to the quality of equipment and to
computerization. In 1957, dependable peristaltic pumps, durable, inert
pump tubing, mirdng coils, and specialized ﬁttings were not available. For
example, in 1957, the inner diameter of mixing coils, ﬁttings, and ﬂow cells
was 3 mm, and ﬂow rates were around 10-15 mL/min, which resulted in
low sample throughput. Sample interaction of greater than 10% was
prevalent, caused mostly by sample residues adhering to the walls of the
tubing and the procedure of using air plugs between samples, instead of the
analyte-free wash that is currently used (7). The precision of
determinations was poor due to variations in the volume of analyte sample
aspirated (7). Severe ﬂuctuations in the ﬂow rate and in the volume of the
analytical stream also exist because of the compressibility of the air
segments as they passed under the pump rollers (8). Sharp peaks without
ﬂat tops were also the norm because of the large dead volumes of the
debubbler and ﬂow cell (8). Improvements in the sampler, debubbler, and
ﬂow cells all took place through the early 1960's to enhance the

performance of the commercial system.

4
3. Enhancing the AutoAnalyzer I

The next major change in the AAI was the double-crook sampler (9).
It resulted in a relatively constant delivery of liquid and air into the
analytical cartridge, because it reversed the role of the sample and air
delivery pump tubes during the sample and wash cycle. Also at this time,
the problem of poor wash in the tubing between the sampler and the
analytical cartridge was addressed by de Jong (10). This problem was
solved by implementing a much more rapid mechanism to cycle the sample
withdrawal tube between samples and by constantly renewing the the
supply of analyte-free wash. Also an improved, low volume ﬂow cell and a

more efﬁcient debubbler were introduced (11).

4. SMA 12/30

In 1965, Technicon introduced the SMA 12130 (12), which was a 12
channel version of an analyzer developed in 1964 (13). This system was
originally developed by Skeggs and Hochstrasser and could perform eight
clinical determinations in parallel, with a rate of 20 samples/hour. It
included a single light source and a photometric detector; these were
multiplexed to eight ﬂow cells on a stage. The steady state signals were

achieved in sequential 20-sec intervals.

5. Second Generation Systems

Second-generation hardware was introduced in 1967 (14). Technicon
produced the SMA 12/60 clinical analyzer, which determined 12 samples in
parallel at a rate of 60 samples/hour (60 x 12 = 720 actual measurements per
hour). This new analyzer contained several key, new hardware

developments. The ﬁrst came about when it was realized that uniform

5

proportioning could be achieved with peristaltic pumps, provided that gas
segments were added in phase with pump pulsations (15). Another major
enhancement included decreasing the internal diameter of the analytical
tubing, mixing coils, and tees, to 2 mm. This reduction in size also
decreased the ﬂow rate of the stream to 2.5 mL/min. Because of the lower
volume of sample coming through the ﬂow cell and its debubbler, the wash
improved signiﬁcantly (14). As a result, the SMA 12/60 was a major
improvement over the ﬁrst generation systems in terms of precision,
analysis time, and cost. All of these new features were introduced in 1970,
in Technicons' AutoAnalyzer II (AAII), which was quickly accepted as the
industry standard (16).

6. Third Generation Systems

Third-generation hardware appeared in 1972 (17,18) in the form of
Technicon's 20-channel SMAC clinical analyzer. It was in this generation
of instrumentation that CFA really matured. It involved pumps, mixing
tees, ﬂow cells, and pump tubing that were further miniaturized, with ﬂow
tubing reaching an internal diameter of 1 mm. These changes led to
improvements in ﬂow rate (< 1 mL/min) and segmentation frequency (from
30 bubbles/min to 90 bubbles/min). Also, dispersion occurring in
unsegmented areas of the CFA system, such as the area between the pump
and the mixing tee, was reduced by using both the peeked sampling
technique and bubble-through ﬂow cells (18). Fiber optics and interference
ﬁlters were introduced at this time and used to multiplex the large number
of ﬂow cells to a single detector. Artifacts from air-segments were removed

with computer technology, and data were taken only from the ﬂat region of
a peak. These changes resulted in a high sample frequency (150

6

samples/hour), low-dispersion system. The demands on sample and
reagent amounts were also decreased. Theory (19,20) predicts the
possibility of building lower dispersion systems with smaller tubing, lower
flow rates, and increased segmentation frequencies. However, the
resultant gains in speed and decreases in sample and reagent consumption
would not offset the technical difficulties and costs associated with further
miniaturization (21). As a result, any new CFA hardware development
past third generation is not foreseen. What will improve is data analysis
and manipulation techniques, and any new systems being developed will
be designed and constructed using existing technology.

It should also be noted here that "in-house" third-generation systems
were also being developed at this time. These systems were being built ﬁrst
by N eeley et al. (22-25) and then later by Patton and Couch (26-28). The CFA
system designed by Patton (27), and later marketed commercially by
Alpkem as the Rapid Flow Analyzer, serves as the model for the system to
be developed and described in this thesis. Details of the system are
discussed later in Chapter 2.

C. Theory of Continuous Flow Analysis

1. Introduction

A generalized CFA system is shown in Figure 1-1 (a typical set-up).
It operates in this manner. Samples are aspirated into ﬂow tubing,
pumped through the tubing into the mixing tee, from the mixing tee
through a mixing coil, and then through a detector to waste. The air bubble
artifact can be removed in two ways, either before or after the sample

detection. If the bubble is to be removed prior to detection, a

0%
.H A

one» <

canon

o Bowman. a

9n... n m

13

$5

doéom

 

 

 

 

nouooaom

 

 

338%:

 

 

 

 

 

 

 

nowuooom

 

 

 

paowmmm

 

 

 

398mm

 

 

 

 

 

390
oEnsm

 

880.89

 

 

 

 

 

 

 

s4

8

debubbler is used. This piece of equipment looks physically like an "F"; the

sample is pumped through to the detector while the air bubble is diverted to
waste. The second alternative is to remove the signal from the air bubble
electronically after the detector by using a bubblegate. An analyte-free
wash solution is aspirated between samples. The volume of sample drawn
into the manifold is dependent on the residence time of the tube aspirating
the sample and the ﬂow rate of the tube. Both sample and wash slugs are
segmented with air bubbles as they enter the mixing tee. After the
measurement has been made, the sample is pumped to waste.

The theory of CFA is well understood, as evidenced in articles by
Snyder and Adler (19, 29), Walker (30), and Thiers (31). Ideally, in the
absence of dispersion, the signal output would be a train of ﬂat-topped
peaks. Transitions from the ﬂat part of the peak (steady-state) to the
baseline would be instantaneous, and the signal from the steady-state
measurement would be proportional to the analyte concentration. The
actual observation, however, is that the transitions to and from the steady-
state condition are skewed. These are rise and fall curves and can be seen
in Figure 1-2. The experimental curve is made up of two contributing
factors, the longitudinal (exponential) and axial (near-Gaussian or
Poisson) contributions to dispersion. Each of these factors is discussed

below.

2. Longitudinal Dispersion

Longitudinal dispersion occurs in the unsegmented zones of CFA
systems. Such zones would be the sample lines (before the mixing tee), ﬂow
cell debubblers, and ﬂow cells. In these areas, dispersion from mixing

effects is dominant and the deviations from the ideal square-wave signals

.38.. $8 35559.... as: 2:. as .58 <5 6083.38 as: 2:. 3 .3 2%:

 

 

 

 

 

3 25a.
om on 3. cm on On a
W . . . . _ H6-
m o
.. do
3.50 SE 1 N6 W
i. co m.
0:9 u
o ﬁn .m I. m
... no
.. so
35m 385 1. he
3
Al tea.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

4— mmqmqv

 

Au

10

are exponential (9). Thiers et al. (31) ﬁrst reported a rough version of the
relationship between concentration and time for longitudinal dispersion,
and then Walker et al. (30) reported the formalized version in Equation 1.1.
The theoretical linear relationship is:

11]. (Ass ' At) = 1n ASS ' t/b (1'1)

In this equation Ass» is the steady state absorbance, At is the absorbance of

the peak proﬁle at time t, and b is the exponential factor or slope constant.
Here, b is the time required for the detector signal to change from any value
of At to a value of 0.37At (8). Because of the exponential relationship
between absorbance and time, the sampling times to obtain peak heights
within desired percent steady states can be readily predicted as values of b.
For example, to obtain peak heights within 99% of the steady state, a
sampling time of 4.6b is needed (32). Percent interaction can also be
calculated as a function of sample and wash time. For example, to obtain a
percent interaction of 0.5, the time between samples must be 5.3b (32). This
model assumes that longitudinal dispersion is greater that axial
dispersion.

Longitudinal dispersion can be compensated for in various ways.
These ways include bubble gated ﬂow cells (23,28,33), pecking samplers (34),
or by curve regeneration (27,35). In curve regeneration, the ﬁnal steady-
state absorbance is calculated from Equation 1-2.

Ass = At + b(dAt/dt) (1'2)

ll

Patton has shown both bubble gating and curve regeneration to be eﬁ'ective
for reducing the effects of longitudinal dispersion (27). Curve regeneration,
in particular, allows peak heights to more nearly approach steady-state and
to reduce sample plug interactions. As a result, the peak proﬁles more

closely resemble ideal peak shapes.

3. Axial Dispersion

The other important contribution to peak shape comes from axial
dispersion. In CFA, mixing between segments is prohibited by air bubbles,
while mixing within segments is enhanced by bolus ﬂow (29). Thus, the the
only mechanism for intersegment analyte transfer is by the stagnant liquid
ﬁlm left on the walls of the ﬂow system after each successive sample and
wash. The concentration of analyte molecules in any segment is

approximated by Equation 1-3 (36).
Ck/c0 = e-q qk/k! (13)

where Ck is the analyte concentration in segment k (k=integer), Co is the
concentration of the initial sample (k=0), and q is a dimensionless
parameter which is the ratio of the volume of liquid that wets the walls of
the entire ﬂow system (Vf) to the volume of a single liquid segment (V8).
When considering a CFA system, the sample is a series of analyte-
containing segments, so the concentration distribution is represented by the
summation of Equation 1-3. For a large number of segments, the initial
Poisson distribution evident by the right-hand side of Equation 1-3 closely

approximates a Gaussian distribution. Thus, axial dispersion is

conveniently expressed as the standard deviation, 0t , of the Gaussian

12

curve. Equation 1-3 was the initial attempt to model axial dispersion, and
unfortunately, fails to relate axial dispersion to experimental variables
such as residence time, ﬂow rate, segmentation frequency, and inner tube

diameter.

4. Snyders Equation

In 1976, Snyder and Adler developed an extensive model of axial
dispersion that expresses q in terms of experimental variables and thus
allows (It to be related to these variables (19). The derivation also allows Vf
to be expressed in terms of the length and diameter of the ﬂow system.
Snyder and Adler (36-39) also extended previous models by accounting for
previously unaccounted for slow mixing between segments and liquid ﬁlm
(29). In 197 6, Snyder expressed his axial dispersion equation in a form that
shows the effects of major experimental variables (20). The model makes

the following several assumptions (20):

1. Air segment volumes are the minimum required to totally occlude
a tube of a given inside diameter ((7/24)1rdt3 = 0.92dt3).

2. The CFA system is perfectly wetted.
3. Longitudinal mixing within the ﬁlm is negligible.

4. The ﬂow system is free from longitudinal dispersion (is. mixing
effects).

Snyders' equation is shown in Equation 1-4.

F ' f 1
a 3 % 7 9!
2 538d,3<F+o.92dtn> 11/3 + y 235(F+o.92dfn) 3n’fst
11

(1-4)
"/3
t

 

 

Q
a
II

 

 

7%1“ Dw’25 a 7%Fd

 

 

13

where:
at = the standard deviation of the peak (s)

(it = the internal diameter of the ﬂow system (cm)

F = the liquid ﬂow rate (mL/s)

n = the segmentation frequency (Hz)

t = the residence time of the sample in the ﬂow system (s)

n = the viscosity of the liquid (poise)

y = the surface tension of the liquid (dyn/cm)

Dw35 = an empirical diffusion coefﬁcient that pertains only to

diffusion in coiled tubes (cm2/s) (20)

The assumption that the ﬂow system is perfectly wetted in insured by
adding a surfactant, such as Brij 35, to the ﬂowing stream. It is also
obvious that the standard deviation of the peak varies directly with the
square root of time, and, as a consequence, sample residence time should
be kept to the minimum time necessary for the reaction to go to completion.
The effect of ﬂow rate, internal diameter of the flow system, and
segmentation frequency can be visualized by plotting the standard deviation
as a function of segmentation frequency for different values of ﬂow rate and
internal diameter, while time is held constant. An example of such a plot
is given in Figure 1-3, and many other examples of such plots are easily
found (20,21,40). The important conclusions to draw from these plots is that
for any ﬂow rate, optimum values of inner diameter and segmentation
frequency exist where the dispersion is a minimum, and, because the
minima are relatively ﬂat, variations in ﬂow rate or segmentation
frequency, by as large as a factor of 4, only affects the dispersion by about 5%
(20). It should be noted that, from Figure 1-3b, it appears that by simply
decreasing the ﬂow rate, even lower dispersion values could be obtained.

However, to do this, the inner diameter must be decreased, and the

14

SN «mm 5 mm 98 853:3 .888
.0328 $8de .oomwuonmmw
no 8893 Boa as» we uSmemv

3533 can .33 3o: .moaoououm

 

oomuwuooﬁmg mo 38am 2E. .wmé 05mg
#3 moooououm oomuﬁooﬁwom
N ma «4 wd v.0 o
_ u _ . .
so «d u :c i

 

(as) noisredsrq

$3 3893.5 acmaouqoﬁmom

8H om cw ow om

— _ _ - u _

so no new

#3 3853.5 oomuﬂooaom
ow mm on

_ p _ — _ — h

EU H.O "umv

 

 

mm om ma 3 m o
_

O>®I>CDIDVUDNH

NH

3

ON

(as) uorsredsiq

(as) uotsxodsrq

Dispersion (82)

Figure 1-3b.

15

 

 

 

6 _
5 _ F=0.0042 ml/s
4 —
F=0.0083 ml/s

3 q
2 _ / F= 0.0167 ml/s
1 .—
0 l l I

0 5 10 15

Segmentation Frequency (s'l)

The effects of segmentation frequency, ﬂow rate, and internal
diameter of the ﬂow system on dispersion. For this plot, (11; = 0.1

cm. All other variables are as found in Ref. 27.

16

segmentation frequency increased. Thus, practical constraints such as
back pressure, ﬂow-cell volumes, and clogging will most likely limit new

eﬂ'orts to reduce dispersion in these systems (41).

D. Comparable Flow Reversal and Flow Recycle Systems

1. Introduction

As the research in this thesis pertains to ﬂow reversal and ﬂow
recycle continuous ﬂow systems, it is necessary to review the current state
of affairs in Flow Reversal techniques (42-53). Currently, the major
emphasis on ﬂow reversal techniques has been when coupled with Flow
Injection Analysis (FIA). For example, Betteridge and Wade have been
employing ﬂow reversal ﬂow injection techniques to manipulate
concentration gradients for either equilibrium-based speciation or
optimization studies (44,47,49). Toei has developed a method for the
determination of glucose in clinical samples by ﬂow reversal FIA (51-53),
while Valcarcel and coworkers have been undertaking the study of

multidetection with a single detector (42, 43, 45, 46, 48, 50).

2. Flow Injection Analysis

To be able to talk about these ﬂow reversal FIA systems, a short
introduction to the principles of FIA is in order. In Figure 1-4a, a typical
CFA set-up is shown and compared to the usual FIA system (Figure 1-4b).
The average FIA system consists of an injection valve, a reaction manifold,
and a detector. In the FIA system, a nonsegmented carrier stream is
pumped continuously through a reaction manifold and then through a
detector. Samples are introduced into the carrier stream (typically)

through the use of a six-port valve. This six-port valve has a sample loop,

Sacha 5E3 $508.88 a 5 Soon: advance 2% 5 3908 mm
3983 05 833 a: 8m 5m H.833 < 3V .oESom woo .Eomwcm #3. new? as com 55 33%... < 3 .YH warm

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

a:
£60 mﬁﬁz uoowoom
n m
A .8838 .m cacao
333 on. a
madam
3
£80 wcmxﬁz .
magnum
_ _ m.
A ~38qu
came? on. llll\/\<<< d uoommom
. 5.

 

 

 

18

which determines the amount of sample introduced into the system. After
injection, the sample and the carrier stream are mixed and sent to the
detector. Thus, the sample, at injection, is a well deﬁned sample plug, but,
as it travels towards the detector, the sample mixes with the carrier stream
and dispersion occurs. This results in peak shapes for FIA typically being
skewed Gaussian, as opposed to CFA peaks having ﬂat tops. The
dispersion in the FIA system is due to the internal geometry of the tubing,
the ﬂow proﬁle, and the molecular dispersion (8).

3. FIA Systems for Concentration Gradients

Betteridge and Wade have initially studied several ﬂow reversal-type
conﬁgurations (44,47,49). These studies involved the use of one or more
pumps, only one detector and one or more valves. One conﬁguration
involved a system that performed a single reversal without a reaction; they
found that the heights of the peaks as a a function of time dropped in an
exponential manner. This is explained by the fact that, in their
conﬁguration, the plug had to travel further backward than forward to
reach the detector. A second configuration involved multiple reversals
without a reaction, with this conﬁguration differing from the ﬁrst by
adding a valve and removing a pump. The resultant peaks were tending
toward Gaussian, with the peak heights slowly decreasing. Finally, a
single reversal with reaction was performed to demonstrate the
applicability of such a system. The proposed system (49) was to be used to
manipulate concentration gradients for equilibrium-based speciation or

optimization studies.

19

4. Multidetection Systems

Toei developed a flow reversal FIA system for the determination of
glucose in clinical samples (51-53). His system is a multidetection, open
ﬂow stream conﬁguration and involves the use of only 1 pump and l
detector. This makes his apparatus relatively simple. In this
conﬁguration, samples are passed through a detector and then pumped
back to the same detector in the reverse direction by use of the same pump.
The ﬂow direction is reversed by employing the use of a six-port valve. In
this conﬁguration, at least two pieces of information were obtained from
each sample. The amount of dispersion was not found to be signiﬁcantly
greater than in the typical FIA system. The effect of several experimental
parameters, such as the effect of flow changeover time, the mechanical
stability of the system, the effect of reaction time, reaction temperature,
sample volume, reagent volume, and the ﬂow-rate of carrier stream were
studied. One problem encountered was increased baseline noise during the
ﬂow changeover due to the rotation noise of the valve, but this noise was
deemed inconsequential because it did not affect the actual analysis. Toei
also found that peak areas were very reproducible, and as such, should be

suitable for kinetic determinations of analytes using two-point assays.

5. General Flow Reversal Techniques

The group of Valcarcel and coworkers in Spain has been very active
in the investigation of ﬂow reversal FIA techniques. Their emphasis has
been on using ﬂow reversal techniques for multidetection in kinetic-based
determinations. They have previously investigated the possibility of
obtaining dual peaks by injecting unusually large sample volumes (54), by

splitting the sample plug and then recombining the streams prior to

20

detection (55), or by using two injection valves located in series, parallel, or
coupled internally (56-58). The more recent work of Valcarcel et al. has
been in studying multidetection in open systems (42,45), and these are the
systems that are more applicable to the flow reversal/ﬂow recycle CFA
systems described in this thesis.

The initial ﬂow reversal FIA system was ﬁrst described in 1985 (42)
and was designed using principles ﬁrst demonstrated by Jensen et al.
(59,60) and Hooley and Dessy (61). This system involved the use of one
detector, one pump, and one switching valve. A second valve was added to
perform sample injections as necessary. The conﬁguration was evaluated
by using three systems. One was without a chemical reaction, one
monitored the disappearance of a product, and one monitored the
appearance of a product. The inﬂuence of typical FIA variables (ﬂow rate,
reactor length, etc.) as well as the inﬂuence of chemical variables
(pertinent to the chemical system being studied) were determined.

Several other applications of this conﬁguration or simple variations
of it have been demonstrated. In one instance, they used a ﬂow reversal
FIA system to determine the viscosity of water-miscible samples (48). In
other cases, they used this system to perform individual and simultaneous
determinations (42,45), to do speciation studies (62), and to calculate the

reaction stoichiometries (46).

Chapter 2

Instrumentation, Software, and Data Processing Techniques

A . Overview

In this chapter, the design of the CFA system is described.
Particular attention was given to the physical descriptions of the various (4-
port and 6-port) valve conﬁgurations, with special emphasis placed on the
6-port conﬁgurations. Also included is a brief description of the individual
parts of the CFA system, as well as manifold design considerations. (For
more detailed information of individual design and construction, consult
the Ph.D. dissertation of Patton (27)). Finally, a discussion of the computer

techniques written for the system is presented.

B. Set-up

A general diagram of a reversal/recycle apparatus is shown in
Figure 2-1. It consists of a pump, a manifold, two switching valves (one
four port and one six port), a source, a double beam detector (calorimeter),
some type of debubbling system (either physical or electronic), and power
supply. Components are described in detail later in this chapter.

The system is operated in a typical CFA manner. The sample, buffer
and reagent(s) are introduced into the system through immersion of the
pump tubes into vials of solution and pumped into the manifold. Once in
the manifold, the sample, buffer, reagent(s), and air are mixed and sent

through mixing coils. After a period of time in mixing coil 1 (the time is
21

22

determined by the length of the coil and the ﬂow rate), the sample is
pumped through the colorimeter and now, unlike typical CFA systems, is
pumped into a holding coil 2 (see Figure 2-1). Once all the sample is
pumped through valve B and into the holding coil, the ﬂow is reversed by
switching valve A and is thus sent back through the colorimeter for another
measurement. This reversal procedure is continued until the desired
number of measurements are obtained or the indicated number of reversals
are accomplished.

In the recycle conﬁguration (see Figure 2-2), valve B is switched after
the last sample has gone through. Now the stream is being propelled by the
buffer coming through valve B. Valve A is switched immediately after the
switch of valve B, and thus the sample ﬂows through valve A and into coil 1.
Once all the sample is completely in coil 1, both valves are switched again.
The sample is now propelled by the buffer stream entering through valve A
and proceeds through the colorimeter again. The whole process then
repeats itself as many times as indicated by the user.

The complete arrangement of equipment for the flow reversal/ﬂow
recycle apparatus is shown in Figure 2-3. Helium gas is used to turn the
valves quickly, and an Opto—22 is used to convert and then relay the TTL
signal from the RTI data acquisition board to the solenoid/actuator.
Descriptions of all other components are given in this chapter. Detailed
valve conﬁgurations are shown for flow reversal in Figures 2-4 and 2-5,
while valve conﬁgurations for ﬂow recycle are shown in Figures 2-6 and 2-
7. Conﬁgurations are given for both valve A being a 4 port and a 6 port
valve, with valve B always a 6 port valve. Currently, the system is operated
on the two 6 port conﬁgurations.

23

 

 

 

 

 

 

 

Reagent

 

 

Buffer

 

Pump

Buﬂ'er

 

Blank

 

 

 

 

 

 

|
I
I
I
Coil 2 Coil 1 '
I
I
I

 

 

 

 

 

Figure 2-1. The general schematic for the Flow Reversal Continuous Flow
system.

24

 

 

 

 

 

 

Reagent

 

 

 

 

Buffer

 

Pump

Buffer

 

 

Blank

 

 

 

 

 

 

Coil 2 Coil 1

 

 

 

 

 

 

 

Figure 2-2. The general schematic of the Flow Recycle Continuous Flow
system.

25

.2333.

Mug
3258

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

390 baasm
.3... I g

 

 

 

 

H
W .8393 oﬁomoogaﬁgom 05 mo
m 30809888 gaogmﬁ 23. Ava warm
w _ u3u=3<
EB
3335
N moo
Suﬁsm
Eamon m 33>
I
\ _ / / 3:me w
_ / ooamuomom d
9
0 \ m / /
was I m.
0/ \ 1 I I /
$3 H :8 I /
K I 1\ ‘ 1&0
\ £30 .833 E I/ '
mom
8&895 J

 

 

 

__
_-

r

||||||||| V
Enmmm oaﬁwm

 

 

 

 

 

 

 

 

 

.333 3 cm 339» mania @395. :4 ”3oz Shadow—3m
mm m 039, .nosﬁswmaoo 18.82: on... an and... 83.5 48.8%: a .89.? sausage. oz? 05 28am 3 can?
gamma 35mg 2% 952%. Am 633 9:56 a BB 989v a 53 58 395,3 3.5 you 55339—8 249» can. .YN warm

 

 

 

 

 

m3;
II»
_
» _
_ _
_
.8
_m
% N
_m
D
m. .mm
8 _.M
ml
:9,0
7m
\‘

 

 

3

 

 

 

 

m2;
ll»
_
v
_ _
_ 1 u
_ w rm
w .m,
J m .m
.m n.
a 3 _m
.8. W
m _m
_ 7w
\‘

 

 

vacuum

0 O
E; .
C 0

Au

.333 3 cm 83a.» music 958.8 =< "3oz
.scosﬁm mm m 25? .qomaﬁsmwcoo $396» as» 3 3524 .3893 a Baa 3335338 33> oﬁ 932% 3
£33 .8589 23$ 0:» 283m Ad .829, ﬂog NE 85 :33 «£0 332,3 Bow .8.“ $3359.38 33> 2E. .m.N mun—ME

 

 

 

 

 

 

 

 

 

 

> >
_ .
+ _ _
_ _ _
_ _
88¢qu
_ .8 .8
7 Tamw _W
2 _H _I _1
a ) )
.J .J I.
_m D m _m 0 O m
u o. .3 u m m _u
.0 U 3 0 I ._.l 9
a Z _.M a T. 3 TM
.mu. m. _$ m.
.m. .m. m. _m
K .A
\‘ _( \ _(
<c>~a> <o>i>

 

 

 

0 98a f 038“ a i
~ m4 3 m a

28

.333 3 cm 839, manic mg ad 882 .353 0.8
82? £3 .Eﬁﬁnmmdoo £993 on» E awﬁ 35o Z .2982 a Eda comedy—sumac". «35> on... 28%. 3 £33
5?ch 35%: on... 36% 3 .313 989w a can 925% a 515 $5 3932 Boa .8.“ 333%8 31> 2E. .c-m 9:55

 

 

 

 

 

 

 

 

 

m3;
II» II»
_ _
_ + _
_ _ _
_ _ _
5035
.8 , .8
_m _ rm,
M 5m N
J: I. .m.
D 1 _m I
m Q u :m
z .m 5 ,m
_o s _p
:4 M. a
0
:W _ 7w

 

£95m A
3 Q

.333 3 cm 829» magnum mkobw =< “3oz
.53 32? £3 .aoﬁwuswwdco £932 23 E 3:... 8502 .2932 a “can maoﬂgammgo 029, 2% 225m 3
can? :33ch BEE an... chgm 3 .339, anon Hum 23 ﬁts «ED 3962 Boa new 53338 33> 23. Sum 953m

m 038/ m 33>

 

 

 

 

 

 

 

 

 

 

 

b J
_ ‘ L
f ~300qu # 8 38¢qu — 8
_ w m _ m
19
m m a. m
_ m H _ 1 mad
m n m
5 u 5 m
_ S w. rmw m.
_m. m. E. W
O
< 955 < 33> .o

 
 

038mm a

3 Q

30

C. Components

1. Pump

The pump used in all experiments was the Model IP-12 variable
speed, 12 channel peristaltic pump from Brinkmann Instruments
(Westbury, NY). It was modiﬁed in-house by replacing the standard 8
roller assembly with a 16 roller assembly (27). The pump motor speed is
controlled by a two decade digital control that provides for linear motor
speed adjustment, as shown in Figure 2-7. The two decade controller has a
range of values from 00-99. A nominal pump setting of 44 was needed for
experimental flow ratings to agree with rated ﬂow values for Technicon
SMA Flow-Rated pump tubes.

Technicon SMA Flow-Rated pump tubes were used in all the
experiments. Flow rates for the pump tubes at various pump settings are
easily calculated through a delivery factor determined for the individual
pump and the ﬂow rating of the pump tubing. It is typical to operate at the
nominal pump speed and change tubing diameter, rather than change the
pump setting. A table giving the ﬂow rate of various tubings at the nominal
ﬂow rate can be found in Table 2-1. Each piece of ﬂow tubing has plastic
shoulder markers at each end indicating tube identity, corresponding to a

certain ﬂow rate. Lubricant is applied regularly to extend the ﬂow tube life.

Table 2-1. Flow rates for various pump tubes at the nominal pump speed

setting of 44.
W2 W4)

orange-black 0.015
orange-red 0.030
orange-blue 0.050
orange-green 0.100
orange-yellow 0.160
orange-white 0.230
black-black 0.320
orange-orange 0.420
white-white 0.600
red-red 0.800
grey-grey 1.000
yellow-yellow 1.200
blue-blue 1.600
green-green 2.000

2. Addition of the Air Bubble

There are two methods that can been used to add air segments to the
ﬂowing stream (27,63). The ﬁrst method was developed by Habig et al. (63).
and is shown in Figure 2-1. It uses two pump tubes (1 and 2) connected in
such a manner that tube 1 acts as a compressor for tube 2. Tube 2 is then
connected to the air inlet of the manifold. Thus, identical segments of
compressed air are trapped between the pump rollers and when the
forward roller leaves the pump platen, the air rapidly expands into the

manifold and creates the air bubble.

32

Delivery Factor

 

 

 

0 I I I I I

0 Z) 40 so 8) 100
Pump Setting

Figure 2-8. A plot of pump tube delivery factor versus pump setting,
indicating that linear pump motor speed adjustment is
possible. The delivery factor is the value by which the nominal
pump tube ﬂow rate must be multiplied to estimate actual ﬂow
rate.

The second method was developed by Patton (27). It involves the use
of an optical chopper, which used information from an opto-interrupter to
add gas from a low pressure gas tank in sequence with the movement of the
pump rollers. This method of phasing used a solenoid valve and the
connection between the gas tank and the manifold. Advantages to this
method include a reduction in the number of required pump channels and
air segments can be added in phase with every second, third, or a forth,

roller lift-off.

33

In this work, air segments are added using the ﬁrst method. This is
because the method is simple and has been proven reliable, and optimal
ﬂow rates for tubes 1 and 2 had already been determined to be 0.05 and 0.1
mL/min (27).

3. Colorimeter

A double beam, ﬁber optic photometer was designed, built and
characterized for the miniature CFA system by Patton and a detailed
description has been given (27). A block diagram of the photometer is
shown for reference in Figure 2-8.

The light source for the detector is a miniature tungsten-halogen
lamp. The source is then connected to the front windows of the ﬂow cells by
a ﬁber optic. This connection is accomplished through the use of
rectangular, black Delrin© blocks that slide along grooves in the base of
the photometer. The ﬂow cells have a pathlength of 1.0 cm, an internal
diameter of 0.05 cm, and are made with sapphire windows. Three-cavity
interference ﬁlters are used to permit wavelength discrimination. These
ﬁlters are 1.27 cm in diameter and have approximately an 8 nm bandwidth.
Easy removal, installation, and proper optical alignment is assured
through holes drilled to a depth of 0.2 cm in the face plate. A

The detector circuitry is encased in a metal cover. The inside
circuitry consists of two silicon photodiodes to detect the radiation and to
convert the radiant power to photocurrent. Current-to-voltage conversion is
accomplished through the use of operational ampliﬁers. Variable gain
ampliﬁers are used to amplify the output from the current-to-voltage

converters. Channel 1 and channel 2 signals are connected to the inputs of

34

.8393 430 m3» 5 wow: uSoSBoAQ Suﬁ .Swonéﬁscc 05 he 89%an 052533 4 .m-m warm

 

 

\

 

meow—928°C
02m

 

 

Banana
2.5313

336325

 

.833
annoyance—J

 

 

 

 

VI.

 

 

 

ovomwgcsm

 

 

£3 35

 

 

I aomos $1311 I

you—E
3539—35

35

a log-ratio ampliﬁer for direct readout of absorbance. Signal outputs are
made available through BNC connectors mounted on the back of the

detector housing.

4. Bubble Gate

An electronic bubble gate is used to remove the unwanted portion of
the signal that .is caused by the passage of the air segment through the
detector. It has been described both electronically and physically by Patton
et al. (28). This bubble gate uses the ﬂuctuations in the detector signal that
are caused by the change in refractive index as the air and liquid segments
pass through the detector to synchronize the time of data acquisition to
include only those times that the liquid segment completely ﬁlls the ﬂow

cell.

5. Manifold Components

Manifold components such as tee's, mixing, coils, connectors, etc.,
were constructed locally. Some pieces had been made previously, while
new parts were fabricated using expert advice (9, 27, 64). All the mixing
tees and four inlet mixing components were made previously (27), and
allow for zero dead volume connections between sample, air, and reagent
tubes and the manifold. However, new mixing and sample coils were made
recently.

The mixing coils were made out of polyethylene tubing. The
procedure used was similar to the process pioneered by Amador (64).
Various lengths of polyethylene tubing were wrapped around 3.0 cm glass
lengths and their ends taped off by masking tape. The rod/tubing

combination was placed in boiling water for approximately 5 minutes and

36

then placed in cold water for 1 minute. After the ﬁnal minute, the
combination was allowed to dry and then the resultant coil removed from
around the glass rod. These coils have a number of advantages when
compared to glass mixing coils. These advantages include cost, durability,
ease of construction, ﬂexibility, and compatibility with ﬁttings that allow for
zero dead volume connections to be made between the coils and tees. The
disadvantage of these coils is that they are less wettable than glass coils,
which results in bubble-break-up due to back pressure created in the CFA
system. However, this problem can be overcome through the use of a

surfactant such as Brij-35.

6. Valves

The valves used in this work are 6-port valves mounted on pneumatic
actuators. The actuators are connected to solenoids, which use helium gas
to turn the valves. Helium, as opposed to compressed air or argon, is used
because the valves need to be turned quickly and helium was determined to
switch valves the quickest.

7. Other Connections

Some connections between tubes cannot be made directly. For
example, short pieces of Teﬂon tubing were inserted into the ends of the
pump tubing for use as sample, reagent, and buffer withdrawal lines. The
pump tubing was not always of small enough diameter to provide for a
secure ﬁt, so plastic sleeves were used to better the ﬁt between the outer
diameter of the Teﬂon tube (typically 0.05 cm) and the inner diameter of the

pump tube. Also, connections between the manifold ﬁttings and the

37

delivery ends of the pump tubing were improved in this manner, as needed.

Worn pump tubes were saved for just such purposes.

D. Sample Introduction

In CFA, samples pass through the pump before they reach the
manifold. In this time period before reaching the manifold (and the point of
insertion for the air bubble), the sample undergoes mixing and dispersion
(because of the laminar ﬂow conditions) between either successive samples
or between sample and wash solution. The intersample air segment (IAS)
that is introduced into the line when the line is switched between sample
and wash serves as a small barrier to intersample mixing. Further
reduction of the intersample mixing and dispersion can occur if the sample
line is "peeked". Pecking involves the quick removal and insertion of the
sample line three or four times when exiting the sample solution and
entering the wash solution and vice versa. Patton has shown this to be

extremely effective for lowering the mixing within a sample (27).

E. Data Acquisition and Processing

Data were acquired through the use of an IBM PC (8088) computer.
A program was written to control the Continuous Flow Reversal/Recycle
apparatus, using the initial parts of a data acquisition routine written
earlier in this laboratory (65,66). The program is called DBEAM. It was
written in a "user friendly" format, which prompts the user to enter all the
necessary parameters for an experiment. It also allows for the correction
of incorrectly entered parameter information. A block diagram showing

the "logical thought progression" of the routine is shown in Figure 2-9.

38

 

 

 

 

 

Reversal
or
Recycle
Reversal Recycle
Parameters Parameters

 

 

 

 

 

 

 

 

I General Parameters

 

 

 

 

I Execute Experiments

 

 

I Real Time Display

 

 

 

I Data Analysis I

 

Figure 2-10. The "thought process" of the CFA Flow Reversal/Recycle
program.

39

The initial choice presented to the user is "Reversal or Recycle"?
Based on the users response, the user is taken through a series of screens
that are aimed at obtaining the parameters necessary for the experiment.
The reversal or recycle parameters ask the user for parameters that are
speciﬁc to the technique, while the general parameters obtain parameters
that would be the same for either experiment. After all the variables are
approved by the user, the experiment is started on the indication of the
user. A real time display presents information to the user that is indicative
of the experiment. After all the data have been collected, the user is given
the option of examining the data in graphical form.

1. Speciﬁc Experimental Parameters

In the ﬂow reversal conﬁguration, there are only a few parameters
needed to be known that are speciﬁc to this experiment. They are 1)
number of samples 2) number of reversals and 3) the desired threshold
(parameter used to determine whether a peak has passed through the
detector, and discussed in greater detail later in this chapter). From this
information, the total number of peaks expected to pass through the
detector is calculated so that the software can end the experiment after all
the necessary information has been obtained. These values are
summarized and are correctable before the user actually begins the

experiment.

40

In the ﬂow recycle conﬁguration, there are a few more variables that

have to be entered. They are as follows:

 

 

Number of recycles

Number of samples

Inner Diameter of mixing coils
Length of mixing coils

Number of inputs into the manifold
The ﬂow rate of the pump tubing
Threshold

99’9”!“pr

 

 

Again, the number of peaks expected through the detector is calculated.
Also calculated is the length of time it will take for the sample(s) to go
through mixing coil 2, completely through valve A, and into mixing coil 1.
After all the sample is in mixing coil 1, then valve A can be switched. The

expression used for calculating this time is shown in Equation 2-1.

Recycle Delay Time = ((100 x CoilLength)/((FlowRate x 3.1415) x ID2)) (2-1)

As usual, all these variables are user correctable before the experiment is

initiated.

2. General Parameters
The general parameters give the user a chance to set variables that
would be common to either experiment. These might include a delay time

before data acquisition begins, solution pH, ionic strength, length of

41

experiment, or the detector channel information, i.e. two sample channels,
one sample, one reference, single beam, etc., among others. This screen

also allows the user to determine the rate of data acquisition.

3. Execute Experiment
This is simply one screen that allows the user to. initiate the
experiment. Data collection begins either immediately or after some user

entered, preset time delay period.

4. Real Time Display

The real time display is a chart that presents the current status for
all the important experimental information. Aspects such as the sample
and reference voltage and the number of reversals or recycles left in this
experiment are updated after the acquisition of each point. A sample
screen is shown in Figure 2-10. Also note that their is a line at the end of
the chart that tells the user the current status. Lines that appear here

include:

 

 

Waiting for the start of peak # X.
Determined the start of peak # X.

In the delay period before valve switch...
Finishing the baseline to point # Y...

P9P!"

 

 

5. Data Analysis
At the current time, the software can generate a number of plots.

The one used most frequently is the absorbance versus time plot, which

 

42

allows the user to immediately view his data after the run is complete.
Unfortunately, the software, at this time cannot adequately calculate the
area under a selected region of the curve, or give the peak maximum of a
selected peak from a series of peaks. It can however, give the peak
maximum for each peak. Also at this time, the user is asked if he/she
wishes to save the data to ﬂoppy diskette. If the response is yes, ﬁles are
written to the ﬂoppy diskette that include both the data points and relevant
run information and parameters. These data ﬁles can be imported into

other statistical packages for numerical analysis.

| 0 l3 i t u 0 A i 3U L ‘ OWAiﬁlm‘R “I: 0
EBQGBAM
I; i 0 8|; . 3 adult. ', t. _ ts”:
Method Reversal
Number of Samples 3
Number of Reversals/Recycles 2
Analyte Signal (V ) 3.00
Background Signal (V) 5.00
Number of Points collected 123
Number of Points expected 451
Number of Peaks to ﬁnd (overall) 9
Number of Peaks Remaining to ﬁnd 5
Number of Reversals/Recycles remaining 1
Present Collection Status In the delay period before valve switch...

 

Figure 2-11. An example screen that might be seen by the user during
an experimental run.

 

43

For this thesis, data ﬁles were copyed onto 3.5", high density ﬂoppy
disks for use in a Macintosh SE/30. The ﬁles were converted to Macintosh
format through the use of a high density Macintosh ﬂoppy drive
(SuperDrivem) and a software program called Apple File Exchange. Once
converted, the data were then entered into a Macintosh
numerical/statistical/graphing program called Kaleidagraphm. This
program allows for standard curve-ﬁtting, user entered curve-ﬁtting
equations, the determination of peak area using selected starting and

ending points, and various statistical functions.

6. Other Programming Considerations
a. Peak Detection

One of the main considerations of this program is how to determine
when a peak has passed through the detector. This problem was solved by
the use of a threshold value (see Figure 2-9). As previously mentioned, the
user is asked to input a threshold value upon which the start and end of a
CFA peak is to be determined. After the ﬁrst 10 points of the run are
collected, the average value of the ﬁrst three points is compared to the
average value of the last three points. That difference is then compared to
the threshold value. If the difference is greater than the threshold value,
then the peak has started. If the difference is less than the threshold value,
then the process is repeated again with points 2, 3, 4 and points 9, 10, 11.
Using the ﬁrst three points and the last three points in a string of ten points
also eliminates the problem of noise causing the start of a peak. If the
threshold is set low, however, this can still be a problem. The other
advantage of this method is that it uses the slope change in the CFA peak to

44

determine the start of the peak, and not just a decrease in voltage. This
helps to discriminate against source drift.

 

 

‘<'~<'~<

 

Absorbance —>

Threshold I I Threshold
xxx - yyy

 

 

Time—p

Figure 2-12. The threshold value is determined using the difference
between the ﬁrst three points and the last three points in a 10
point series. When the threshold is large enough, the
beginning of the peak is "determined". The end of the peak is
determined in the same manner, only the difference must
also be near the initial value used when determining the start
of the peak. The "x" represents the ﬁrst three points and the

y represents the last three points in a series of 10 points.

45

To determine the end, the absolute difference is taken continually
along the CFA peak and compared to the threshold. If the difference is
sufﬁcient, then a second condition must be met. That condition compares
the average of the ﬁrst three values (which would be the last points coming
down the tail of the peak) to the average of the ﬁrst three values that
determined the start of the peak to some preset range. Detecting the tail in
this manner ensures that the peak returns to near baseline, and prevents

premature endpoint determination.

b. Valve Switching

Valves are switched by using TTL signals generated by the RTI-815
data acquisition board at the appropriate times. Each TTL signal is sent to
an Opto-22, which converts the RTI signal to light, and then the light
triggers a relay that operates solenoids, which then control pneumatically

actuated, 6 port valves.

c. Delay Times

The delay time between the detector and valve 2 is a predetermined
number of points. For example, if the last point that determined the end of
the peak was point 115 and the delay time is 15 points, then the valve(s) will
switch after point 130 is detected. This procedure actually gives a 30 point
line between peaks, which provides ample time to get a ﬂat baseline
between peaks. The delay time when doing recycles is actually the amount
of time that will take the sample to go completely from coil 2 to coil 1, as
calculated from user entered variables.

Chapter 3

Characterization of the Flow Reversal/Flow Recycle System I

A. Overview

This chapter covers the characterization and performance of the
Flow Reversal/Recycle CFA system. Experiments investigating the
inﬂuence of different numbers of reversals and recycles on peak shapes
(areas and heights) are described. A phenol red/borax buffer is used as the
sample. Second, the analytical performance of the new system is compared
with the analytical performance of a typical CFA system using the
analytical determination of nitrite by the Griess Reaction (27). Finally, the
interaction or carryover is determined for a multiple number of valve
switches. Finally, the performance of the peak detection routine is

evaluated using data from the Griess reaction.

B. Experimental
1. Reagents for the Dispersion Experiments
a. Borax Buffer
The borax buffer was prepared by adding 3.800 g of borax to 1 L
of distilled water. The pH was adjusted to 9.5 using 50% NaOH. The
solution was then ﬁltered and Brij-35 was added.

46

47

b. Phenol Red Dye

The dye was made from a 5.0 mL dilution of phenol red with
the borax buffer from a 1.0 mM/0.001 M NaOH basic dye solution and then
ﬁltered.

2. Reagents for the Nitrite determination
a. NaN 02 stock solution
0.1725 g of dried NaNOz was diluted to 500 mL using distilled
water. The solution was ﬁltered and then, from the stock solution, 2.5, 5.0,
10.0, 15.0, and 20.0 M dilutions were made. These solutions were stored in

glass bottles and refrigerated.

b. Sulfanilamide (SAN)

SAN was prepared by adding 5.000 g in 250 mL of water, plus
100 mL of concentrated HCL in a 500 mL volumetric ﬂask. The HCl is
added to attain the proper solution pH. The solution was ﬁltered, and then
one milliliter of Brij-35 was added and the solution was stored in an amber
bottle and refrigerated.

c. N -(1-Naphthyl)ethylenediamine dihydrochloride (NNED)
NNED was made by the addition of 0.5000 g of NNED diluted to
500 mL using distilled water, ﬁltered, and then stored in an amber bottle

and refrigerated.

48
C. Instrumentation

All instrumentation, computer control and data acquisition
techniques are the same as previously described in Chapter 2. Speciﬁc
pump conﬁgurations and manifold designs~ will be detailed during the

introduction to each experiment.

D. Peak Shape Determinations

1. Introduction

After the construction of new instrumentation, it is necessary to
characterize the system. Because the performance of CFA systems have
been previously well characterized in regards to the effects of ﬂow rate, coil
length, etc., on peak shape, and the new system is essentially a traditional
CFA system with valves inserted into the ﬂow path, it follows that those
same effects should hold for the ﬂow reversal/ﬂow recycle system. As a
result, the experiments performed were primarily concerned with the eﬁect
of valve switching on the peak shape. .

The ﬁrst set of experiments were performed using a phenol red/borax
buffer dye system that absorbs at 540 nm. Sample and wash times were 20
and 20 seconds, respectively, as recommended by Patton (27). Pecking
during sample introduction was used to help minimize dispersion across
the pump. The pump and manifold conﬁgurations are shown in Figure 3-
1. The conﬁgurations were the same for both the ﬂow recycle and the ﬂow
reversal experiments, with only the connections to the valves being changed
(as described in Chapter 2). Trials were run for 1 sample and multiple
valve switches, two samples and multiple switches, and ﬁnally three
samples and one valve switch. The results were used to determine what

effect the valve switches had on peak area and peak height.

49

 

 

 

 

 

 

 

 

 

 

 

 

Air Orange-Green
Orange-Black
Phenol Red Black-Black J >
Borax Orange-Green
Pump

Figure 3-1. The pump conﬁguration and tubing selections for the
manifold used to study the effect of valve switching on peak
shape using a phenol red/borax buffer system.

50

2. Inﬂuence of Valve Switching

At the outset of this project, there was concern as to what effect
switching valves was going to have on a ﬂowing stream. If the switching
caused turbulence problems it would signiﬁcantly degrade the peak shape
and peak area (signiﬁcantly degraded being roughly a 5-7% difference
between the original peak and the peak detected after a switch). This
degradation would cause the current project to have signiﬁcant problems.
Fortunately, this was not the case. The inﬂuence of valve switching on
peak area and peak height is shown in Figures 3-2 and 3-3. From these
ﬁgures, it can be concluded that valve switching does not signiﬁcantly
change the peak shape in terms of peak height and peak area. Therefore,
this type of method could be useful for analytical work. Examples of output
from the reversal/recycle apparatus are given in Appendix A.

Examining these two ﬁgures closely, one can see that, as with typical
CFA systems, the peak area and peak height change the longer the sample
is in the system. This is as predicted by Snyder's model. The fact that the
peak area is changing both in the positive and negative directions indicate
that the sudden change in pressure is having some sort of effect on the
stream, most likely compressing or expanding the air bubbles, which
would make the sample segment of the ﬂowing stream either spread out or
condense. As such, the area of the peak either rising or falling, depending
on whether or not the bubble is being expanded or condensed.

Consider ﬁrst peak height from Figure 3-3, peak height changes by
approrn'mately 3% over the course of 4 to 5 valve switches. These variations
can be due to a number of instrumental parameters, most notably source
drift. Also, notice that the increase in peak height appears to level off
around 3% after the fourth reversal or recycle, which would indicate that

51

this variation in peak height is probably a limitation of the instrumentation.

Dispersion values for the reversal and recycle peaks were obtained
using a graphical estimation method (9,27). They are presented in Table
3.1. The values do not follow a linear increase with time, probably due to
the momentary change in ﬂow rate and pressure introduced into the
system during a valve switch. These changes make dispersion prediction
difﬁcult, as evidenced by the difference between values predicted by
Snyder’s equation. The important aspect of the dispersion values is,
however, that dispersion is less than three seconds, with is within

acceptable limits.

Table 3-1. A comparison of calculated and experimental dispersion
values. All dispersion values are in seconds.

Valve Switch Reversal Recycle

Number Theory Calculated Theory Calculated

0 5.79 0.960 5.79 1.071

1 ‘ 7.70 2.627 14.24 2.367

2 9.58 2.368 19.27 L600

3 11.07 1.975 23.09 1.593

4 12.45 2.394 26.34 1.965

5 13.80 2.772 — —
Average 10.06 2.183 17.75 1.719

52

Another intriguing possibility for the reversal/recycle system would
be if it could be used to analyze more than one sample per run. That is, a
user could introduce two peaks into the system and get multiple
information on multiple peaks during the same run. Preliminary
investigations into this possibility yielded encouraging results. The tricky
aspect of this situation is to remember that in ﬂow reversals, the last peak
through the detector is the ﬁrst peak back into the detector when the ﬂow is
reversed. With recycles, the peaks always enter the detector in the same
order. Preliminary data for this type of analysis are given in Tables 1-4,
located in Appendix B.

53

 

 

 

5
c... —I— Recycle
o
4 - - A- - Reversal
:: ca
.2 a)
4.3
g 2
ca. '5 3
as 5': .
no \
Q Q \
DD 2 \
0 ﬂ \
u: < \
O 1 5
8°
0

0 1 2 3 4 5 6
Number of Reversals or Recycles

Figure 3-2. The change in peak area as a function of average peak area after
each successive reversal or recycle.

Average Peak Height

% Change as a Function of

Figure 3-3.

54

 

 

 

5
—l— Recycle

4 - - A- Reversal
3 i " ' ' ' ' ' ‘
2
1
0

0 1 2 3 4 5 6

Number of Reversals or Recycles

The change in peak height as a function of average peak height aﬁer
each successive reversal or recycle.

55

E. The Griess Reaction

1. Introduction

In order for the ﬂow reversal/recycle system to be preferred over a
regular CFA system, in needs to preform as well or better during analytical
determinations. Because of the large amounts of data that can be collected
during a reversal or recycle (when compared to regular CFA, it can easily
be 3 times more), it may be more precise. For example, in regular CFA,
runs are done in triplicate and, as such, three data points are obtained. In
reversal or recycle CFA, runs are done in triplicate, but also with two
reversals or recycles each time, and, as a result, nine data points are
obtained. However, it will also take longer, which is a disadvantage. The
amount of sample required will be equal (assuming 3 trials), but the
reversal/recycle system will use more buffer. The possibility exists to do 1
trial with two reversals or recycles to obtain three points, but to do this
would be risky because something could be wrong with your ﬁrst sample,
and the analyst would get three values for the same "bad" peak and then all
of the data would be wrong without the user even realizing the mistake.

The reaction chosen to test the reversal/recycle system is the Griess
reaction (27). It is shown in Figure 3-4, and produces a red dye that absorbs
at approximately 550 nm. The manifold set up is shown in Figure 3-5.
Notice that it uses a ten-tum coil to mix the sample and SAN before NNED
is added. A sample time of 20 s and a wash time of 20 s were used, and
runs were made in triplicate.

The Griess reaction allows us to examine several aspects of the
reversal/recycle system. Using the Griess reaction, a calibration curve is
constructed and is compared to the working curve generated using the

standard CFA technique (from now on, CFA implies standard CFA, while

56

reversals or recycles will be speciﬁed). In addition to the working curve, an
interaction test pattern is obtained by sampling, in this order, 2.5 uM, 20
uM, and 2.5 uM, in accordance with the method of Theirs et al. (67).

Evaluating the peak widths for differences, and evaluating the time
between the end of one peak and the start of the next peak after a valve
switch (and thus the consistency of our peak detection routine), are also

done at this time.

. H=1
N02 + H2N_©" SOzNHz —p—> +Nz—QSOZNH2

SULF

SOzNHg

WFQW. . _.._... I?
HNCHZCHZNHZ -
NNED 00

NHZCHZCHZNH

Figure 3-4. The reaction scheme for the determination of NOg' using the
Griess Reaction.

57

 

 

Air Orange-Green I

 

 

Orange-Black

 

10 turn coil
Smple Orange-White I

I 1M ’

 

 

SAN Orange-Black

 

 

NNED Orange-Black

 

 

 

 

Pump

Figure 3-5. The pump conﬁguration and tubing sizes for the manifold
used for the Griess reaction.

2. Calibration Curves

The ﬂow reversal/ﬂow recycle system performed reasonably well, but
failed to give as high quality of results as did CFA in terms of %RSD. A
reason for this might be because the valve switching is having some sort of
effect on the peak that is not obvious physically, or is not illustrated through
the dye experiments. If the valve switches are having an effect, perhaps by
compressing a peak at one time and then letting it expand the next, it would
be reasonable to imagine the %RSD could be greater than that for CFA. In
addition, because there are more points being taken, there is more of a
chance for this error to occur and be detected. One more explanation
would be that with the small degradation in peak shape associated with
reversals or recycles, and increase in %RSD could be expected. Percent

relative standard deviation data are given in Table 3-2, using 2.5 ppm for

58

comparison. Also given in the table is an example of the within run
variations for reversals and recycles. The values under the reversal and
recycle columns in the %RSD overall row represent the %RSD for the entire
nine peaks (3 trials x 3 peaks each trial) available. The values under the
reversal and recycle columns represent the %RSD between the three peaks
in each trial.

The standard curves were plotted for both peak area and peak height,
and the results are given in Table 3-3. Despite the %RSD's favoring CFA,
the plots almost perfectly overlay each other. Therefore, reversal/recycle
methods have proved an adequate analytical technique to at least a

sensitivity of 2.5 ppm.

Table 3-2. A comparison of % relative standard deviations for 2.5 ppm
nitrite solution using CFA and ﬂow reversal/ﬂow recycle CFA.
The CFA data was obtained using three trials.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Reversal Recycle
CFA Triall Tria12 Trial3 Triall Tria12 Trial3
%RSD
4%
Peak
Area 0.71 2.89 2.39
(considering all trials) (considering all trials)
Peak
Height 0.02 1.03 0.67
@511
mm
JEN;
Peak
Area NA 0.72 4.81 0.81 2.34 8.10 1.98
Peak
Height NA 0.18 0.02 0.03 0.02 0.00 3 x 10-4

 

 

 

 

 

 

 

 

59

Table 3-3. The ﬁgures of merit from the calibration plots of CFA, Flow
Recycle, and Flow Reversal using the Griess reaction.
Calibration plots are for both the peak area and the peak height.

QEA W W
slope 0.615 0.0327 0.623 0.0322 0.620 0.0322
y-intercept 0.134 0.00288 0.100 0.00267 0.103 0.00228
rxy 0.998 0.999 0.998 0.999 0.998 0.999
3. Percent Interaction or Carryover

Because of the number of times a sample is being reversed or
recycled, the amount of sample interaction or carryover that might be
occurring is very important. This phenomenon occurs as the ﬂowing
stream leaves behind sample along the wall of the tube, and all or a portion
of the sample is picked up by the next sample plug. Thus, the original
sample is decreased from what it should be, and the following sample is
increased more than expected. In a reversal or recycle system, this is
deﬁnitely of major concern.

As mentioned before, samples of 2.5 uM, 20 11M, and 2.5 uM nitrite
were run and the percent interaction was calculated by taking the
difference between the maxima of the two shoulder peaks, and then
dividing by the height of the center peak and multipling by 100.

Results from the interaction test patterns were very encouraging.
They are given in Tables 3-4 and 3-5. The percent interaction values are

comparable to known CFA systems (27), and do not seem to increase after

60

each valve switch. A sample interaction output is given in Appendix II for

both reversals and recycles.

Table 3-4. The percent interaction for four ﬂow recycles.

. 0.085450
2 20.0 0.645190 0.149
1 2.5 0.086410
1 RECYCLE
1 2.5 0.086410
2 20.0 0.645%0 0.132
1 2.5 0.085560
2 RECYCLE
1 2.5 0.085568
2 20.0 0.645250 0.155
1 2.5 0.086567
3 RECYCLE
1 2.5 0.086610
2 20.0 0.645230 0.161
1 2.5 0.085572
4 RECYCLE
1 2.5 0.085575
2 20.0 0.645210 0.167
1 2.5 0.086650

61

Table 3-5. The percent interaction for four ﬂow reversals.

. 0.085442
2 20.0 0.645190 0.156
1 2.5 0.086448
1 REVERSAL
1 2.5 0.086451
2 20.0 0.645200 0.129
1 2.5 0.085621
2 REVERSAL
1 2.5 0.085621
2 20.0 0.645190 0.132
1 2.5 0.086472
3 REVERSAL
1 2.5 0.086501
2 20.0 0.645140 0.132
1 2.5 0.085650
4 REVERSAL
1 2.5 0.085652
2 20.0 0.645190 0.135
1 2.5 0.086525

4. Evaluation of the Peak Detection Routine

All of the nitrite calibration data that has been generated offers a
good opportunity to evaluate the peak detection and valve switching routines
in the the software that have been written. Recalling that in the peak
detection routine it isn't necessary to know exactly when a peak comes
through, just a rough estimate, it is interesting to note just how consistent
the peak detection routine is at picking the beginning and the end of peaks
and keeping a roughly equal length baseline between them. The data are
displayed in Figure 3-6. It shows that the majority of the time, the detector

keeps the distance between reversal peaks around 38 seconds, which is

62

correct knowing that wash times are generally around 20 seconds, and
there is a 20 point or 10 second delay period between the end of a peak and
the valve reversal. It is also correct for the recycle peaks to be around 160

seconds between peaks.

‘ Reversals
I Recycles

separated two peaks

 

Number of times a length of time

 

0 42.5 85 127.5 170 212.5 255 297.5 340
seconds

Figure 3-6. A plot representing the frequency of time from the endpoint
detection of one peak and the detection of the front of the peak
following a valve switch.

Chapter 4

Analysis of the Reaction Rate of Glucose using the Trinder reaction to
Investigate the Possibility of Flow Reversal or Flow Recycles being used for
the Determination of Reaction Rates.

A. Overview

One of the possible uses of the reversal/recycle system is for the
determination of rates of reactions, or kinetics. The problem with current
CFA systems is that there is only one point obtained per sample, unless
multiple detectors, pumps, split streams, etc. are used. By being able to
reverse or recycle a ﬂowing stream, more than one point per sample is
obtained. In fact, if the reaction is slow enough, the whole reaction
progress curve should be able to be determined. The system developed here
has this capability, and it has the advantage of needing only 1 pump and 1
detector. Thus, it is very cost effective, because it doesn't need additional
pumps or detectors, and requires very little sample.

This chapter describes the efforts to prove that the reaction rate of
glucose with glucose oxidase can be determined with reversals and
recycles. The reaction was ﬁrst performed using standard, single-beam
spectrophotometric equipment because it is a competing method. The

results of both experiments are discussed.

63

B. Introduction

The reaction used to compare the reversal/recycle CFA to another
analytical technique needs to be slow. Enzymatic reactions are prime
examples of reactions whose rate can be easily controlled. The reaction
chosen here is the conversion of B-D-glucose to D-glucono-d—lactone, or
more commonly known as the glucose oxidase reaction. Because the
product is not detectable via visible light methods, a detection reaction is
needed to form a colored product. The reaction chosen was the Trinder
reaction, which has been described previously in the literature (68-71). The
product formed is a pink-red dye, which is detected at 510 nm. The reaction
is presented in Figure 4-1. The manifold arrangement is shown in Figure

4-2.

C. Experimental

All solutions were the same for both the spectrophotometric
experiments and the reversal/recycle experiments. The ratio of the Trinder
reagent to the amount of glucose introduced into the system was kept the
same throughout; however, the volume of sample was not the same. The
experiments used 3 mL of composite reagent and 1 ml of glucose, while the
Reversal/recycle system used the same reagents in an approximate 3:1

ratio by employing the proper ﬂow tubes.

1. Buffer

The buffer used was 0.01 M potassium phosphate solution, with a pH
of 6.85. This was determined to be the optimum pH for this reaction (71).
The solution was ﬁltered, Brij-35 added, and the solution was stored in a

plastic bottle.

65

2. Glucose Solution

A 1000 ppm glucose stock solution was made by mixing 0.5005 g of
glucose with distilled water in a 500 mL volumetric ﬂask. The solution was
then ﬁltered. Working solutions of 5, 10, 15, 20, 40, 60, and 80 ppm were
made using the appropriate dilutions from the stock solution.

3. Glucose Oxidase Enzyme Solution

The enzyme solution was prepared by mixing 0.06 g of glucose
oxidase in 50 mL of distilled water. It was made fresh daily and stored at
5 °C.

4. 3,5-dichloro-2-hydroxyphenyl sulfonic acid (DCPS)
A 10 mM stock solution was made from 1.02 g of solid in 0.5 L of
distilled water. The solution was ﬁltered, stored in a ﬂask, and

refrigerated.

5. 4-aminoantipyrine (AAP)
A 10 mM stock solution of AAP was made from 1.330 g of solid in
0.500 L of distilled water. It was ﬁltered, stored in a stoppered ﬂask, and

refrigerated.

6. Composite Reagent

The composite reagent was made by combining 20 mL of buffer, 10
mL of the glucose oxidase enzyme solution, 5 mg of peroxidase, 5 mL of
AAP, and 5 mL of DCPS. Brij-35 was added, and then the solution was
stored in a stoppered ﬂask and kept out of room light.

.5583 538.8% Swath. 2.5 was .8308." 63238 885% 25. AA. ouawwm

wdmmm + cum”. +

 

 

 

m<< mmOQ
MEN—v
2 «mo £0 £2 6
\ 6
01.
ﬂ mmUlz +
2 6333.89 / O + «03%..
w o 6 0 mo
wagon—-méaoogwd mmoogmdé
mo _ mo _
_ SO @3338 \_ mo
«Cam + C I NO +
98on /
ol.. 0 mo ol: 0

mommo momma

67

 

 

 

 

 

 

 

 

 

 

 

 

Air Orange-Green
Orange-Blue
Composite Orange-White >
Reagent
Sample Orange-Green
Pump

Figure 4-2. The reaction manifold and pump tubing choices for the glucose
oxidase/Trinder reaction.

68

D. Instrumentation

Again, all the reversal/recycle instrumentation is the same. Sample
and wash times are 20 seconds each. For the alternate single beam
method, the Heath instrumentation was used, and was operated at a slit
width of 2 mm and a wavelength of 510 nm. The source was a tungsten
lamp. Data taken while using the Heath apparatus were recorded on a

strip chart recorder.

E. Results

A relative rate of reaction versus concentration plot for the data
obtained from the Heath apparatus is given in Figure 4-3. It was obtained
by measuring the rate increase of absorbance over time, after an induction
period of approximately 30 seconds. This induction period is important,
because it ensures that the rate is being measured during approximately
the same time period in both techniques. If the induction period is not
considered, the rate for the Heath determinations would be determined in a
time frame different than that of the reversal/recycle system, and it
wouldn't be a fair comparison. From the plot in Figure 4-3, it is obvious
that the linear range for this reaction is between 1 and 50 ppm. It is not
linear above 50 ppm because, at high concentrations, enzyme saturation
occurs. For this reason, and because any concentration over 20 ppm is
kinetically too fast for the reversal/recycle system, only concentrations less
than or equal to 20 ppm were used.

The reversal data are very interesting in terms of appearance, and
illustrate a problem that will exist when doing kinetics with recycle CFA.
Examining Figure 4-4, which is the result of a trial using 5 ppm glucose,
notice the shapes of the peaks. The ﬁrst and third peaks are exactly as is

69

expected, that is they are ﬂat-topped peaks. The second and fourth peaks
are different, and this is because of the ﬁrst in, last out principle of
reversals. With the initial peak, all of the sample volume passes through
the detector at exactly the same relative time in terms of when the reaction
started. When the sample reverses and enters the detector for the second
time, the points on the back of the curve are now on the front and are
analyzed ﬁrst. The points that were on the front of the curve (initially) are
now on the back of the curve, and as such have had a longer time to react.
Because of this time discrepancy, the peak is no longer ﬂat-topped. It rises
from left to right, indicating that the sample towards the back of the sample
plug has had more time to react, and thus has a greater absorbance. This
is reﬂected in peak numbers 2 and 4. Interesting to note is that peak
number 3 is also ﬂat-topped, but only because the time discrepancy evens
out during this reversal. The gives the overall appearance of a ﬂat-topped
peak, a spike peak, a ﬂat-topped peak, and then another spike peak.

70

 

 

0.005 -
E 0.004 —
3
,g 0.003 "
5%
.9 Cl
:5 0.002 -
3
‘3 0.001 — C]
0 5 I T I I I I
0 10 Z) 30 40 50 a)

Glucose concentration (ppm)

Figure 4-3. The rate of the glucose oxidase reaction using a single beam,
Heath instrument.

The recycle data are exactly as expected, giving ﬂat-topped peaks
each time because all parts of the sample plug pass through the detector in
the time frame during the course of the reaction. This is evidenced in
Figure 4-4. An interesting feature in this plot is that the analyst can
actually see the reaction slowing down, as evidenced by the peak heights not
increasing in a linear fashion after 500 seconds.

To obtain rates for the reversal and recycle experiments, the average
peak height and average time during that peak height were obtained for
each peak and then plotted. This is illustrated in Figure 4-6 and uses the
absorbance versus time data from Figure 4-4. The slope of the linear
regression line was taken to be the rate for that concentration. Once the

rates for each concentration was known, a relative rate versus

71

concentration curve could be generated. The regression for these lines, as

well as for the Heath experiments, are given in Table 4-1.

Table 4-1. A comparison of the calibration curves for the glucose
oxidase/Trinder reaction. The data reﬂect the linear
regression ﬁt to the rate versus concentration data. The units

are L ppm'1 for the slope and ppm for the y-intercept.

Reversal Recycle
Heath Measured Adjusted Measured Actuated
slope 1.01 x 104 2.60 x 104 1.13 x 10-4 2.64 x 10-4 1.15 x 104
y-intercept 0.0000114 0.000228 — 0.000196 —
rxy 0.98 1.0 — 0.99 —

The results from the reversal and recycle system agree well, but both
of the experimental rates are appronﬁmately double the rates obtained using
the Heath instrument. This is because the ratio of the composite reagent to
the glucose sample is different due to limitations in the selection of pump
tubing sizes. 'Once the rate is adjusted for this difference, the rates are
similar. There are two possible explanations for the differences between
the adjusted reversal/recycle data and the Heath data. One possible
explanation for this could be mixing effects. To mix the glucose with the
composite reagent during the Heath experiments, a syringe was used.
This syringe served two purposes. First, it provided an easy way to inject
small volumes of glucose into the spectrophotometric cell. Secondly, by

quickly drawing solution back into the sample and then reinjecting it

72

several times, turbulent ﬂow and mixing is accomplished. This is much
better than using a magnetic stir bar, which causes large amounts of noise.
In CFA, the mixing is accomplished by coiled tubing inserted after all the
components have been mixed together, which provide adequate time'for the
reaction to go to completion, but also allow for mixing to occur. Secondary
ﬂow (bolus ﬂow) is present, and that has been”, shown to signiﬁcantly
increase radial mixing (29). The differences in rates suggest that CFA
provides for increased mixing over the syringe method, and thus a
corresponding increase in the reaction rate is seen. Another possible
explanation for this could temperature differences. .During the
. reversal/recycle experiments, the reaction temperature was kept at room

temperature, but not controlled by a water bath.

73

0.25

0.185 [M

 

Absorbance
O
E3

 

 

 

 

 

 

0.055 ﬂu L.

—0.01

 

0 50 100 150 Z!) 50
Time (s)

Figure 4-4. An example of the data obtained from glucose oxidase/Trinder
reaction using ﬂow reversals.

74

0.6

0.4475 F‘

Absorbance
o

0.1425

 

 

 

 

 

 

 

n J I

0100200300400500600700800
Time(s)

 

-0.01

Figure 4-5. An example of the data obtained from glucose oxidase/Trinder
reaction using ﬂow recycles.

75

0.25

0.2

0.15

0.1

Absorbance

0.05 D

 

 

0 40 80 120 160 200
Time (s)

Figure 4-6. The resulting absorbance versus time plot from the recycle data in
Figure 4-4.

76
F. Summary

From the data presented in this chapter, it has been shown that
periodic ﬂow reversals or recycles in a CFA system can be used to obtain
rates of reactions. This was shown by using the glucose oxidase reaction,
coupled with the Trinder detection reaction. Unusual peak shapes were
obtained in ﬂow reversals, but can be explained by examining the times at
which the absorbances were measured in relation to when the part of the
sample stream is passing through the detector. Peaks for flow recycles
were ﬂat-topped, as expected. Rate data were obtained from the average
peak heights (absorbances) and the average time during the measured peak
height, and reasonable rates were obtained. Results were compared to a
single beam, Heath instrument, and they varied by about 10%. This
difference can be explained by variations in temperature and mixing

efficiency.

Chapter 5

Future Projects

A. Overview

The ultimate goal of this project is to create a fully functional,
computer-controlled, ﬂow reversal/ﬂow recycle continuous ﬂow analysis
system. The system will be able to perform, with just a few modiﬁcations in
manifold arrangement, coil length, and pump tubing sizes, a variety of
regular CFA, CFA Reversal, and CFA Recycle experiments. Also available
will be a variety of curve ﬁtting possibilities, statistical functions, and
routines to do kinetic analysis. The advantages of such a system would be
its inexpensiveness, fleidbility, durability, small solution volumes, and ease
of usage. It would be applicable to a wide range of slow, kinetic reactions
that are not suitable for analysis by such methods as stopped-ﬂow or FIA.
These chemical systems are of current interest in this research group and
are common to such areas as enzyme reactions or food analysis.

The following projects need to be worked on to accomplish the

aforementioned goals.

B. Valve Design
To make ﬂow reversal or ﬂow recycle accessible without changing
the valve connections, a new valve conﬁguration has been designed that

will allow both reversals and recycles to occur. This new valve

77

78

conﬁguration adds a third, 6-port valve to the existing two valves, as shown
in Figure 5-1.

The connections during a ﬂow reversal valve switch are diagrammed
in Figure 5-2. The flow of the sample is directed through valve a, coil 1, the
detector, and then valve b sequentially, as in previous examples (see
Chapter 2). After all of the sample has passed through valve b and in coil 2,
the flow is reversed by switching only valve a. As a result, the ﬂow is
reversed back through valve b and into the detector. The ﬂow should be able
to be reversed in this manner several times without signiﬁcant peak shape
deterioration. Similarly, in Figure 5-3, ﬂow recycles can be accomplished
by switching valves a, b, and c simultaneously. This will direct the ﬂow
through valves c and a, and then into coil 1. Once all the sample is in coil 1,
all the valves are switched again, and the sample goes through the detector.
A large number of recycles are expected to be available without a large
change in peak shape.

The software already written for the flow reversal/ﬂow recycle
apparatus provides for the choice of either reversals and recycles in the
same program shell, and it would be very convenient if the user could do

either method without having to change valve connections.

C. Computer Enhancements

Several improvements in the current system are foreseen. These
improvements include, but are not limited to, real-time, graphical output, a
full range of curve analysis capabilities, and kinetic routines, in addition to

more user friendly utilities.

79

 

 

 

 

 

Reagent

Buffer

Buﬂ'er

Blank

 

 

 

 

 

Pump

 

 

 

 

 

 

 

 

_________J

 

 

I —

 

+-—-I

 

 

Valve B

Figure 5-1. The general schematic of the Flow Reversal/Flow Recycle

continuous ﬂow system. In this conﬁguration, connections
do not have to be switched when choosing between ﬂow
reversal or flow recycle modes.

80

.333 3 cm 829, music ESE“ =< "302 .bwncmagm 93 O 93 m 82.; .acﬁausumaoo 3956." 2: E
.35 c332 .3326.” a .8»? 833%8 93> 23 952% 3 338 dogma 333 one 952% An
.nmmmov «£3 oﬁohuéamugom 05 we ncmawusmmcoo Hagan Bow 2: you 53926980 31> 2E. .m-m E

m 028’

 

 

 

Z 11°C)

/———————————>

(51110 9105033 103) 193mg

 

3

 

 

 

meg;
0’. I. II
.. .
O O —
.\
+ _
_ m _
.38.: u _
- .8
_ _m
.m .m.
x m 0...... a.
_w H. 0’ h
a T. o o _0
_s ‘. 9.
AW 0 o _m.
:w .\ /_./A\

 

Aw

81

.333 3 am 333 «£338 285 :4 "3oz .3253 0.8 $2? =u 333399.38 2983 23 E
323 8302 .2992 a pods 5393938 025, 23 9323 3 can? 833% 333: 23 9523 Ad
.ammmov 02? Seneca—«830m 23 .3 8395338 3982 Be: 23 he 5325.338 31> 2E. .m-m 0.53m

 

 

 

 

 

as? as? .l
\O O I I > O O I II >
o a. _ .0 o _
I. .\ _
+ _ + _
_ _ O
_ m _ m. _
88309 _ Z uSoSoQ 7... _ H
_ H
_ , m. _ _m
5.. .m, J. 1...).
m. _ ma _ m... 0 U 33> m
_. o. 9.... _. m... m. . _...
7m 0 \. a 7.. I I. «A.
. .. . .. a... _s .. J...
Em ‘ m. . .
am . . _m _.W . .\ _m.
.m. I. .00. .m.
... / 2...... .. /

 

 

 

 

mﬁﬁwwa maﬁam

82
1. Output Screens

One feature currently being investigated is making possible the
choice of output screens during the experiment. Currently, the user can
only monitor the progress through the use of the chart style output (Figure
2-10). The output format planned would allow for the user to choose either
the chart or a real-time, graphical representation of the CFA peak as it
passes through the detector. Also, it would be possible for the user to
actually be able to cycle between the chart and the graphical output styles

whenever desired.

2. Graphing Enhancements

Plans are for the user to be able to select from a wide range of plots,
such as a simple voltage versus time plot, a straight absorbance versus
time plot, an absorbance versus absorbance plot or various log-log plots.
These options will allow the user several different ways to explore

experimental data.

3. Peak Analysis

In this area, plans are being made to make peak measurements
available immediately after the experiment, as opposed to having to convert
data for use in another graphing/statistical program. Needed
measurements include peak area, peak height, and a graphical method for
estimating dispersion. The user should be able to select the start and end of
the data values to be integrated, and to specify or have the computer
determine a baseline for reference. The user should also be able to select a

region of the continuous ﬂow signal and expand it further (zoom).

83

Another useful calculation that could be automated would be the
sample interaction percentages. These capabilities would allow for
improved data analysis, and would let the user have some immediate
knowledge of the experimental results. This is in contrast to the present
situation, where data are exported to another analysis routine. Any
corrections or additional experiments that need to be done have to be done

later, which costs the investigator valuable time.

4. Manual Valve Switching _

Another useful feature that should be added to the software would
allow the user to manually control the valve switching. Currently, the
computer has complete control over the valve switching, and the only way a
user can interrupt a run is to abort it. Manual control would allow the user
to switch valves whenever desired, so features such as quicker and/or
slower valve changes or switching in the middle of a peak would be

possible.

5. Computer-Controlled Peristaltic Pump

A computer-controlled pump is also planned for this system. This
would make investigating the effect of ﬂow rates on chemical systems
possible. It would also make stopped-ﬂow CFA, another branch technique
of CFA and Stopped-Flow, possible.

6. Computer-Controlled Syringe Pump
Sample introduction is another area where improvements can be
made. This research lab owns a syringe pump capable of being computer

driven. The syringe pump can deliver small volumes of analyte with great

84

precision. The addition of the syringe pump would allow for very precise
analytical work to be done with any of the CFA system variations that are
available. '

7. Curve Smoothing and Regeneration

Another area of interest is in curve smoothing and curve
regeneration. Patton (27) has shown that these techniques are both useful
and possible. Peaks that have been smoothed or regenerated will more
closely resemble the curves predicted by Snyder's model, and dispersion

calculated from these peaks are very close to predicted dispersion values.

8. Smoothing of Points During Data Acquisition

Currently, the software averages 15 points to make every one point
and it is then 1 point on the CFA curve. All 15 points have equal weight
when computing the average of that one point. Other smoothing methods,
such as the Savitsky-Golay, weight data points in order to their relevance to
the point of interest. Points that are close get a large weight, and points
that are far away get less of a weight. This might give CFA peaks a better

shape and reduce instrumental errors.

D. Segmental Flow Injection Analysis

There has been an interesting paper just published which combines
FIA with CFA (72). The authors have designed a special, sample
introduction valve which allows them to inject an FIA sample plug into a
segmented ﬂowing stream. The result is an FIA plug in a CFA setting,
which gives the advantages of both techniques and eliminates some of the
disadvantages.

85

It would be interesting to to try and adapt their valve conﬁguration
for the sample introduction to the ﬂow reversal/ﬂow recycle system. This
should allow for both FIA reversal and recycle to be done without the
dispersion problems that exist in FIA.

E. Kinetics

A major advantage of doing reversal or recycle CFA is the ability to
obtain several measurements on the same sample. This is in contrast to
normal CFA, which only obtains one steady-state value for each sample
sent through the system. Easily envisioned is the ability to take one sample
and follow the kinetic curve by reversing or recycling the sample several
times. This would allow several points to be obtained for each sample, and
a log plot could be generated using the acquired absorbance versus time
values. Of course, this system would be applicable only when determining

the kinetics of relatively slow reactions.

F. Diode Array Detection

The CFA system should be easily interfaced to the diode array
detector in the Crouch laboratory. The diode array can obtain both
absorption and ﬂuorescence information nearly simultaneously. The
system should be able to monitor mixtures containing several absorbing
species, or mixtures containing both chromophores and ﬂuorophores.
Being able to reverse a ﬂowing stream through the diode array detection
system would allow enormous amounts of information to be obtained in one
experiment. This large amount of information would require the use of

chemometric techniques for data reduction, smoothing, and analysis.

86

G. Light Source Modiﬁcations

The light source currently used is a Tungsten halogen lamp, which
is ﬁne in the visible range, but has some shortcomings when approaching
the UV. Speciﬁcally, it fails to provide enough light to the detector, starting
around 400 nm. In order to do kinetic analysis, frequently a second
reaction is needed to form a colored product because the original product
formed is not colored. It would be useful to have some UV detection
reactions available, such as the conversion of I' to 13', which is detected at
380 nm. To accomplish this goal, several approaches are available. New
diodes that have a larger range could be placed in the detector and a new
source module could be built that accommodates a light source different
from the current tungsten halogen lamp. Finally, different uses of ﬁber
optics could be investigated, in hopes of trying to get more light to the
detector.

H. Other Possibilities

There are certainly a number of other possibilities for this
reversal/recycle system. Several samples could be introduced into the
system at once, and, thus, several experiments done together. Another
possibility would be to introduce a complete working curve and unknown
into the system, and then reverse or recycle the peaks several times. Now,
in the course of one experiment, several repetitive measurements of the
same working curve can be obtained in addition to several measurements of
the unknown sample. Also, standard additions could possibly be
automated with this system. Thus, precision information is easily
measured in one experiment. In addition, the software would be able to plot

the curves and determine the absorbance of the unknown sample.

Appendices

87
Appendix A

The following outputs are given as examples of data for the
reversal/recycle CFA system. Output is given for one phenol red dye

sample and 5 reversals, one dye sample and 4 recycles, and interaction test

patterns for both 4 reversals and 4 recycles.

0.7

0.5225

0.345

 

 

 

 

o Absorbance

.1675

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

UUUUUL

0 50 100 150 200 250
Time (s)

 

-0.01

Figure A-1. The output from I dye sample with 5 reversals.

88

0.7

0.5225

Absorbance
é

 

 

 

 

 

 

 

 

 

 

 

 

 

0.1675
43.01 r “ L L
O 100 $1) 300 400 5(1) 6“) 700 800
Time (s)

Figure A-2. The output from 1 phenol red dye sample and 4
recycles.

89

0.7

0.5225

0.345

Absorbance

0.1675

 

 

 

 

 

 

 

 

 

 

 

JLJULMLHMW

0 1002b3004005006®7m800
Time(s)

-0.01

Figure A-3. The interaction pattern for 2.5 M, 20.0 M, 2.5 uM nitrite
solutions and four reversals.

0.7

0.5225

0.345

Absorbance

0.1675

 

 

 

 

 

 

 

 

 

 

 

-0.01

JLLJlLJl MW

400 m0 800 1000 IZJO
Time (s)

Figure A-4. The interaction pattern for 2.5 “M, 20.0 M, 2.5 pM nitrite
solutions and four recycles.

91
Appendix B

The following data sheets are for multiple phenol red dye samples
and multiple reversals or recycles. It is intended to demonstrate that peak
areas and peak shape are relatively constant and that it could be possible to
introduce multiple samples and then do reversals or recycles with out peak
degradation. These numbers are signiﬁcant to three decimal places.

The start and stop values are the beginning and end of the peaks, and
serve as the limits of integration when determining peak areas. They are

measured in seconds.

92

Table B-1. Data sheet on experiments with 2 dye samples and 2

reversals.

(Absorbance)
PeakNumber Start Sup Start-Stop PeakArea PeakHeight

1 21.5 40.5 19.0 10.595960 0.673980

2 52.5 72.5 20.0 10.189880 0.671460

2 85.5 104.0 18.5 10.240660 0.669580

1 116.0 137.0 21.0 10.533750 0.672928

1 150.5 168.5 18.0 10.170370 0.670080

2 185.0 205.0 20.0 10500690 0.672810

avg 19.4 10.371885 0.671806
std 1.1 0.191810 0.001735

1 41.0 59.5 18.5 10.474760 0.669880

2 73.0 94.5 21.5 10.267410 0.671190

2 108.5 129.0 20.5 10.451450 0.667320

1 148.5 166.0 17.5 10.177910 0.670880

1 183.5 203.0 19.5 10.498650 0.672810

2 219.0 244.5 25.5 10.617030 0.671130

avg 20.5 10.414535 0.670535
std 2.8 0.161683 0.001835

1 25.5 44.0 18.5 10.340980 0.673080

2 54.0 71.5 17 .5 10.093030 0.675170

2 86.5 105.0 18.5 10.088930 0.673860

1 113.5 134.0 20.5 10.558850 0.674600

1 150.0 174.5 24.5 10.198370 0.674740

2 180.5 199.0 18.5 10.259850 0.674340

avg 19.7 10.256668 0.674298
std 2.6 0.177056 0.000738
Total AVG 19.9 10.347696 0.672213
Total STD 2.2 0.176850 0.001436

Table B-2.

PeakNumba' Start

<HN¢O¢DNH

w
on

E

m
<HNODODNH
81m

HNWWNH

avg

Total AVG
Total STD

13.0
43.5
70.5
101.0
130.0
162.0

21.0
48.0
78.0
109.0
135.0
164.0

18.5
52.0
82.0
110.5
140.0
173.0

Shop
30.5

60.0
87.5
121.0
148.5
178.5

37.5
65.5
97.5
126.0
153.0
183.5

38.0
69.0
100.0
130.0
159.0
193.5

93

Data sheet on trials with 3 dye samples and 1 reversal.

(Absorbance)

Start-Stop PeakArea PeakHeight

17.5
16.5
17.0
20.0
18.5
16.5
17.7

1.4

16.5
17.5
19.5
17.0
18.0
19.5
18.0
1.3

19.5
17.0
18.0
19.5
19.0
20.5
18.9
1.2

18.2
1.3

9.861701

10.058830
10.480180
10.527490
10.298790
10.137200
10227 365
0.256646

10.058830
9.861702

10.191700
10.473000
10.297390
10.191700
10.179054
0.207795

10.191700
10.480150
10.058830
10.058830
10.060330
10.164870
10.169118
0.163359

10.191846
0%?

0.66893)
0.67052)
0.6697 90
0.669780
0.671430
0.667410
0.6%653
0.001372

0.670520
0.668990
0.6697%
0.66899)
0.671430
0.669720
0.669895
0.000943

0.66972)
0.669790
0.670520
0.67062)
0.669785
0.669726
0.670027
0.000%

0.669858
0.000913

94

Table B-3. Data sheet for trials with 2 dye samples and 2 recycles.
(Absorbance)
PeakNumber Start ﬁip Start-Stop PeakArea PeakHaidit

1 22.0 39.5 17.5 10.592880 0.674780

2 49. 5 67.0 17.5 10.140680 0.672650

1 214.5 233.0 18.5 10.573630 0.669560

2 245.0 %60 21.0 10.209750 0.673430

1 425.5 446.5 21.0 10.590230 0.669940

2 457. 5 477. 5 20.0 10.190030 0.672580

avg 19.3 10.382867 0.672157

std 1.6 0.223296 0.00%)29

1 59.5 80 .0 20.5 10.249650 0.670060

2 93.5 112. 5 19.0 10.14472) 0.672870

1 268.5 286.0 17.5 10.592880 0.673980

2 296.0 313 .5 17.5 10.124380 0.671560

1 462. 5 483.0 20.5 10.248160 0.670060

2 494.0 514.0 20 .0 10.194540 0.671890

avg 19. 2 10.259055 0.671737

std 1.4 0.171475 0.001549

1 22.5 43.0 20.5 10.206640 0.669890

2 55.0 75.0 20. 0 10.156000 0.672950

1 212. 5 232.0 19. 5 10.217810 0.669580

2 244.5 %35 19.0 10.138940 0.672470

1 409.0 429.0 20.0 10.597550 0.673980

2 436.5 455.0 18.5 10.151670 0.672710

avg 19.6 10.244768 0.671930
std 0.7 0.175717 0.001779
Total AVG 19.3 10.295563 0.671941
Total STD 1.3 0.190163 0.001786

Table B-4.

PeakNumber Start

<WNHOONH

N
N

5i

m
a<wNHCONH
UQ

“NHOONH

avg
std

Total AVG
Total STD

13.0
50.0
86.0
219.5
258.5
294.0

13.0
55.0
92.5
228.5
270.0
314.5

15.0

54.5

98.0
223.0
264.5
302.0

Stop
31.0

66.5
104.5
238.5
280.5
311.5

32.0
71.5
111.0
248.0
288.0
332.5

33.5
72.0
115.0
239.5
283.0
320.0

95

Data sheets for trials with 3 dye samples and 1 recycle.

(Absorbance)

Start-Stop PeakAma PeakHdght

18.0
16.5
18.5
19.0
22.0
17.5
18.6

1.9

19.0
16.5
18.5
19.5
18.0
18.0
18.3

1.0

18.5
17.5
17.0
16.5
18.5
18.0
17.7
0.8

18.2
1.2

9.863906
10.058830
10.485520
9.858550
10.254940
10.475310
10.166176
0.283883

10.533150
10.058750
10.483210
10.4402Z)
10.349320
10.498720
10.393895
0.175962

9.849703

10.480900
10.040630
10.034720
10.360750
10.345290
10.185332
0.244944

10.248468
0.234930

0.668990
0.670520
0.66973)
0.6695%
0.671430
0.668990
0.669873
0.000952

- 0.669010

0.670500
0.669810
0.668800
0.670970
0.668330
0.669570
0.001032

0.669490
0.669790
0.670970
0.670700
0.669670
0.670970
0.670%5
0.000688

0.669903
0.000891

LIST OF REFERENCES

$999289.“

10.

11.

12.

13.

14.

15.
16.

97

LIST OF REFERENCES

Horvai, G. and Pungor, E., Critical Reviews in Analytical Chemistry,
17, 231, 1987.

Skeggs, L.T., Am. J. Clin. Path., 28, 311, 1957.

Skeggs, L.T., US. Patent 2,797,149, June 25, 1957.

Ruzicka, J. and Hansen, E.H., Anal. Chim. Acta, 78, 145, 1975.
Stewart, K.K. and Beecher, G.R., Hare, P.E., Fed. Proc., 33, 1439, 1974.
Stanley, R., J. Auto. Chem., 4(6), 175, 1984.

Thiers, RE. and Oglesby, KM., Clin. Chem, 10, 246, 1964.

Patton, OJ. and Wade, A.P., in Analytical Instrumentation
Handbook, Ewing, G.W. ed., Dekker, New York, Ch. 27, p 979, 1990.

Furman, W.B., Continuous Flow Analysis: Theory and Practice,
Dekker, New York, p 61, 1976.

de J ong, E.B.M., US. Patent 3,134,263, May 26, 1964.

Practical Automation for the Clinical Laboratory, White, W.L.,
Erickson, M.M., Stevens, S.C. eds., C.V. Mosby Company, St.
Louis, MO, 1968.

Whitehead, E.C., in Automation in Analytical Chemistry, Technicon
Symposia 1965, Mediad, Inc., White Plains, NY, p 437, 1966.

Skeggs, LT. and Hochstrasser, H., J. Clin. Chem. (Winston-Salem,
NC), 10, 918, 1964.

Smythe, W.J., Shamos, M.H., Morgenstern, S., Skeggs, L.T., in
Automation in Analytical Chemistry: Technicon Symposia 1967,
Vol. I, Mediad, Inc., White Plains, NY, p 105, 1968.

Smythe, W.J., US. Patent 3,306,229, February 28, 1967.
Lobrig, LB. and Booth, R.L., in Automation in Analytical Chemistry:

geiggécon Symposia 1972, Mediad, Inc., Tarrytown, New York, p

17.

18.

19.

21.

27.

£3

31.
32.

98

Isreeli, J. and Smythe, W.J., in Automation in Analytical Chemistry:
Technicon Symposia 1972, Vol. I, Mediad, Inc., Tarrytown, NY, p
13, 1973.

Diebler, H. and Pelavin, M., in Automation in Analytical Chemistry:
Technicon Symposia 1972, Vol. I, Mediad, Inc., Tarrytown, NY, p
19, 1973.

Snyder, L.R. and Adler, J .J ., Anal. Chem, 48, 1017, 1976.

Snyder, L.R., J. Chronwtogr., 125, 287, 1976.

Snyder, L.R., in Advances in Automated Analysis: Technicon
International Congress 1976, Vol. I, Mediad, Inc., Tarrytown, NY,
p 76, 1977.

Neeley, W.E. and Sing, H.C.,Am. J. Clin. Chem, 61, 840, 1974.

Neeley, W.E., Wardlaw, S., Swinnen, M.E.T., Clin. Chem. (Winston-
Salem, NC), 20, 78, 1974.

Neeley, W.E., Wardlaw, S., Sing, H., Clin. Chem, 20, 424, 1974.

Neeley, W.E., Wardlaw, S., Yates, T., Hollingsworth, W.G., Swinnen,
M.E.T., Clin. Chem, 22, 227, 1976.

Patton, C.J. and Crouch, S.R., Anal. Chim. Acta, 179, 189, 1986.
Patton, C.J., Ph.D. Dissertation, Michigan State University, 1982.
Patton, C.J., Rabb, M., Crouch, S.R., Anal. Chem, 54, 1113, 1982.
Snyder, L.R. and Adler, J .J ., Anal. Chem, 48, 1022, 1976.

Walker, W.H.C., Pennock, C.A., McGowan, G.K., Clin. Chim. Acta,
27, 421, 1970.

Thiers, R.E., Cole, R.E., Kirsch, W.J., Clin. Chem, 13, 451, 1967.

Walker, W.H.C., Continuous Flow Analysis: Theory and Practice,
Furman, W.B. ed., Dekker, New York, 1976.

Habig, R.L., Schlein, B.W., Walters, L., Thiers, R.E., Clin. Chem, 15,
1045, 1969.

Diebler, H., Pelavin, M., in Advances in Automated Analysis,
Technicon Symposia 1972, Vol. I, Mediad Inc., Tarrytown, NY, p 7,
1972.

37.

39.

41.
42.

47.

49.

51.
52.

99

Walker, W.H.C., Clin. Chim. Acta, 32, 305, 1971.

Hrdina, J ., in Amino Acid Analysis: 6th Colloquium, Monograph No.
3, Technicon Corporation, Tarrytown, NY 1967.

Theirs, R.E., Reed, A.H., Delander, K., Clin. Chem, 17, 42, 1971.
Begg, R.D., Anal. Chem, 44, 631, 1972.
Walker, W.H.C. and Andrew, K.R., Anal. Chim. Acta, 57, 181, 1974.

Snyder, L.R., Lavine, J ., Stoy, R., Conetta, A., Anal. Chem, 48, 942A,
1976.

Snyder, L.R., Anal. Chim. Acta, 114, 3, 1980.

Rios, A., Luque de Castro, M.D., Valcarcel, M., Anal. Chem, 57,
1803, 1985.

Valcarcel, M., Luque de Castro, M.D., Rios, A., Spanish Patent, No.
554725, May 1986.

Wade, A.P., Betteridge, D., Crouch, S. R., "Simulation of Flow
Reversal FIA for Rapid Automated Speciation and Optimization
Sudies", FACSS XIII Conference, St. Louis, MO, October 2, 1986.

Rios, A., Luque de Castro, M.D., Valcarcel, M., Anal. Chim. Acta,
179, 463, 1986.

Rios, A., Luque de Castro, M.D., Valcarcel, M., J. Chem. Educ, 63,
552, 1986.

Wade, A.P., Stults, C.L.M., Crouch, S.R., "Flow Reversal FIA--Muses
on Uses", FACSS XIV Conference, Detroit, MI, 1987.

Rios, A., Luque de Castro, M.D., Valcarcel, M., Talanta, 34, 915, 1987 .
Betteridge, D., Oates, P.B., Wade, A.P., Anal. Chem, 59, 1236, 1987.

Bios, A., Luque de Castro, M.D., Valcarcel, M., Anal. Chem, 60,
1540, 1988.

Toei, J .,Analyst, 112, 1565, 1987.
Toei, J .,Analyst, 113, 475, 1988.
Toei, J ., Talanta, 36, 12, 1989.

57.

59.

61.
62.

100

Fernandez, A., Luque de Castro, M.D., Valcarcel, M., Anal. Chim.
Acta, 193, 107, 1987.

Fernandez, A., Luque de Castro, M.D., Valcarcel, M., Anal. Chem,
56, 1146, 1984.

Fernandez, A., Luque de Castro, M.D., Valcarcel, M., Analyst,, 112,
803, 1987.

Ruz, J ., Rios, A., Luque de Castro, M.D., Valcarcel, M., Fresenius' Z.
Anal. Chem, 322, 499, 1985.

Rios, A., Luque de Castro, M.D., Valcarcel, M., Anal. Chem, 58, 663,
1986.

Dahl, J .H., Espersen, D., Jensen, A., Anal. Chim. Acta, 105, 327, 1979.
Espersen, D. and Jensen, A., Anal. Chim. Acta, 108, 241, 1979.
Hooley, DJ. and Dessy, R.E., Anal. Chem, 55, 313, 1983.

Ruz, J ., Rios, A., Luque de Castro, M.D., Valcarcel, M., Talanta, 33,
199, 1986.

Habig, R.L., Schlein, B.W., Walters, L., Thiers, R.E., Clin. Chem, 15,
1045, 1969.

Amador, E., Clin. Chem. (Winston-Salem, NC), 18, 164,1972.
Stults, C.L.M., Michigan State University, unpublished work (1985).
Bowman, L.E., Michigan State University, unpublished work (1987).-

. Theirs,R.E., Meyn, J ., Wilderman, Clin. Chem, 16, 832, 1970.

Trinder, P., Ann. Clin. Biochem, 24, 6, 1969.
Barham, D. and Trinder, P., Analyst, 97, 142, 1972.
Stults, C.L.M., Ph.D. Dissertation, Michigan State University, 1985.

. Thompson, R.Q., Ph.D. Dissertation, Michigan State University, 1982.

Tian, L.C., Sun, X.P., Xu, Y.Y., Zhi, Z.L, Anal. Chim. Acta, 238, 183,
1990.